{"id":1249,"date":"2026-11-01T08:58:00","date_gmt":"2026-11-01T12:58:00","guid":{"rendered":"https:\/\/drugchatter.com\/insights\/?p=1249"},"modified":"2026-09-26T14:02:33","modified_gmt":"2026-09-26T18:02:33","slug":"what-happens-when-investors-ask-ai-about-your-company-for-biopharma-the-answer-is-often-wrong-outdated-or-missing-entirely","status":"publish","type":"post","link":"https:\/\/drugchatter.com\/insights\/what-happens-when-investors-ask-ai-about-your-company-for-biopharma-the-answer-is-often-wrong-outdated-or-missing-entirely\/","title":{"rendered":"What Happens When Investors Ask AI About Your Company? For Biopharma, the Answer Is Often Wrong, Outdated, or Missing Entirely"},"content":{"rendered":"\n
\"\"<\/figure>\n\n\n\n

An investor researching a biopharmaceutical company in 2026 is more likely to open ChatGPT, Perplexity, or Gemini than to read the 10-K first. A March 2026 Investing.com survey of 938 U.S. retail investors found that 62% already use AI tools to inform investment decisions, and 53.5% specifically use chatbots such as ChatGPT for investing-related research.[1]<\/sup> The same survey found that 38.9% of respondents worry about incorrect or misleading AI recommendations, the single largest concern investors named.[1]<\/sup> For a public biopharma, that combination of heavy reliance and open distrust is a specific, documentable business risk, and it converges directly with the drug-information-accuracy problem DrugChatter and DrugPatentWatch already track for patients, clinicians, and competitive intelligence teams.<\/p>\n\n\n\n

The Short Answer<\/strong><\/h2>\n\n\n\n

When an investor asks an AI model about a pharmaceutical or biotech company, the model does three things at once: it pulls from training data that lags real-world regulatory and clinical events by months to years, it weights authoritative sources (FDA.gov, peer-reviewed journals, SEC filings) unevenly, and it synthesizes an answer with the same confident tone whether the underlying facts are current or stale. Independent testing puts hallucination rates for large language models at 3% to 27% depending on model and domain.[2]<\/sup> For a mid-cap or small-cap biotech with thin public-source coverage, the gap between what the AI says and what the label, the docket, or the 8-K actually says can be large enough to shape a trading decision.<\/p>\n\n\n\n

How Many Investors Are Actually Asking AI Right Now<\/strong><\/h2>\n\n\n\n

Retail investors have already normalized it<\/h3>\n\n\n\n

The Investing.com data is not an outlier. Coverage of the same survey found that among retail investors who have tried AI tools at all, chatbots such as ChatGPT are the most commonly used category, ahead of dedicated stock-scoring subscription products.[3]<\/sup> A separate 2026 academic study of 121 individual French investors found that investors primarily use ChatGPT for data analysis, risk assessment, and sentiment analysis when researching positions.[4]<\/sup> None of this is confined to day traders. Institutional adoption is running in parallel: Goldman Sachs has reported AI adoption across roughly 90% of the firm, and Bridgewater launched a dedicated AI-driven fund.[5]<\/sup><\/p>\n\n\n\n

What this means for a public biopharma specifically<\/h3>\n\n\n\n

A biotech’s investor base is unusually dependent on technical, fast-moving facts: patent expiration dates, trial readouts, FDA advisory committee votes, label changes, and safety signals. These are exactly the categories where AI models are weakest, because the underlying facts change on a regulatory and litigation calendar that outruns any model’s training cutoff. An investor asking “when does the company’s key patent expire” or “what happened in the Phase 3 trial” is asking a question with a documented, dated, verifiable answer, and increasingly asking an AI model instead of the primary source.<\/p>\n\n\n\n

The Five Findings That Matter<\/strong><\/h2>\n\n\n\n
    \n
  • 62% of retail investors already use AI to inform investment decisions, and 53.5% specifically use chatbots for research<\/strong> \u2014 Investing.com, March 2026 survey of 938 investors.[1]<\/sup><\/li>\n\n\n\n
  • 22% of Fortune 100 companies now name AI hallucinations, inaccuracies, or misinformation as a material risk in their 10-K, as of November 2025<\/strong> \u2014 EY Center for Board Matters analysis.[6]<\/sup><\/li>\n\n\n\n
  • 72% of S&P 500 companies disclosed at least one material AI risk in 2025, up from 12% in 2023<\/strong> \u2014 a sixfold increase in two years.[7]<\/sup><\/li>\n\n\n\n
  • A single AI chatbot demo error wiped roughly $100 billion from Alphabet’s market value in one trading day<\/strong> \u2014 the closest documented case of an AI-generated factual error moving a public company’s valuation.[8]<\/sup><\/li>\n\n\n\n
  • Independent benchmarking finds hallucination rates of 3% to 27% across major LLMs depending on domain<\/strong>, with specialty and mid-cap biotech topics landing on the high end of that range because of thin public-source coverage.[2]<\/sup><\/li>\n<\/ul>\n\n\n\n

    When AI Got a Public Company Wrong: Four Documented Cases<\/strong><\/h2>\n\n\n\n

    Alphabet, February 2023: a $100 billion answer<\/h3>\n\n\n\n

    In a promotional demo for its Bard chatbot, Google’s own AI stated that the James Webb Space Telescope took the first images of a planet outside the solar system. That claim was false \u2014 the first exoplanet image was captured by the European Southern Observatory’s Very Large Telescope in 2004.[9]<\/sup> Alphabet shares fell as much as 9% the following trading day, erasing roughly $100 billion in market capitalization.[8]<\/sup> This is not a case of a third party misusing AI to defame a company; it is Google’s own product making a factual error about an unrelated scientific topic in a promotional video, and the market treating that error as a signal about the company’s underlying AI competitiveness. The mechanism that matters for other public companies is the same one: an AI-generated factual error becomes a market event once enough people see it and treat it as informative.<\/p>\n\n\n\n

    Walters v. OpenAI: fabricated wrongdoing, real legal exposure<\/h3>\n\n\n\n

    A Georgia radio host sued OpenAI after ChatGPT, asked to summarize a real federal lawsuit, fabricated a claim that the host had been criminally charged with embezzlement in a case where he was not even a party.[10]<\/sup> OpenAI ultimately prevailed because the plaintiff could not establish the fault standard required for a public figure under defamation law, and because the specific user who received the false output recognized it as unreliable.[11]<\/sup> Legal commentators reviewing the case have been explicit that the ruling turned on the facts of that one interaction, not on a blanket rule that AI-generated business or reputational claims cannot be defamatory.[12]<\/sup><\/p>\n\n\n\n

    Wolf River Electric v. Google: a company, not a person, claiming financial injury<\/h3>\n\n\n\n

    A Minnesota solar installation company alleges that Google’s AI Overview told searchers the company had settled a deceptive-practices lawsuit with the state attorney general and, as a result, the company lost contracts and revenue.[13]<\/sup> Unlike the individual-plaintiff cases, Wolf River is a business alleging concrete financial harm from an AI-generated statement about its regulatory and legal standing, which is precisely the fact pattern investor relations teams should have in mind: an AI answer about compliance status or litigation history, stated as fact, reaching prospective customers or investors before anyone catches the error.<\/p>\n\n\n\n

    Air Canada: a company is bound by what its own chatbot says<\/h3>\n\n\n\n

    A Canadian small-claims tribunal held Air Canada liable after its customer-service chatbot gave a passenger inaccurate information about bereavement-fare refund policy, rejecting the airline’s argument that the chatbot was a separate legal entity for which the company bore no responsibility.[14]<\/sup> The ruling is now cited across legal commentary as the baseline precedent that a company cannot disclaim its own AI-generated statements as someone else’s error.[14]<\/sup> The case involved a customer-facing support bot, not investor communications, but the underlying principle \u2014 you own what your deployed AI says about you \u2014 generalizes directly to any IR chatbot, virtual analyst-day assistant, or AI-generated FAQ a public biopharma puts on its own site.<\/p>\n\n\n\n

    \n

    An EY analysis of Fortune 100 10-K risk disclosures found that about 1 in 5 Fortune 100 companies (22%) now flag AI hallucinations, inaccuracies, misleading outputs, misinformation, disinformation, or bias as a material risk, as of November 2025.[6]<\/sup><\/p>\n<\/blockquote>\n\n\n\n

    What the Boards Already Know: The 10-K Evidence<\/strong><\/h2>\n\n\n\n

    AI risk disclosure went from rare to routine in two years<\/h3>\n\n\n\n

    Just 12% of S&P 500 companies disclosed a material AI risk in their 2023 annual filings. By 2025, that figure had reached 72%.[7]<\/sup> A separate analysis found that AI now appears as a standalone risk factor \u2014 its own captioned section, not folded into “cybersecurity” or “emerging technology” \u2014 in 33% of fiscal-year-2025 10-Ks, up from roughly 1% three years earlier.[15]<\/sup> Reputational risk is the single most frequently cited AI concern, disclosed by 38% of S&P 500 companies in 2025.[7]<\/sup> A separate filings-intelligence review found the specific term “hallucinations” now appears in 177 annual reports, and that 3,545 10-Ks filed in 2026 mention artificial intelligence at all.[16]<\/sup><\/p>\n\n\n\n

    The SEC is treating AI statements as ordinary disclosure, not a novel category<\/h3>\n\n\n\n

    The SEC’s first “AI washing” enforcement actions, against investment advisers Delphia (USA) Inc. and Global Predictions Inc. in March 2024, penalized the firms a combined $400,000 for overstating what their AI systems actually did.[17]<\/sup> SEC leadership has been explicit that the same disclosure obligations that apply to any material statement apply to AI claims: companies must have a reasonable basis for what they say about their AI use, and investors must be able to rely on it.[18]<\/sup> That standard cuts in a second, less-discussed direction that matters more for IR teams than for AI vendors: if an AI model is telling investors something false about your company’s pipeline, patents, or safety profile, and your public disclosures do not address that gap, the disclosure controls question shifts from “did we oversell our own AI” to “did we adequately warn investors that third-party AI describing us may be wrong.”<\/p>\n\n\n\n

    Disclosure metric<\/th>2023<\/th>2025<\/th>Source<\/th><\/tr><\/thead>
    S&P 500 companies disclosing a material AI risk<\/td>12%<\/td>72%<\/td>Harvard Law School Forum on Corporate Governance[7]<\/sup><\/td><\/tr>
    10-Ks with AI as a standalone risk factor<\/td>~1%<\/td>33% (FY2025)<\/td>Finrep 2026 practitioner guide[15]<\/sup><\/td><\/tr>
    Fortune 100 companies flagging AI hallucination\/misinformation as material<\/td>\u2014<\/td>22% (as of Nov. 2025)<\/td>EY Center for Board Matters[6]<\/sup><\/td><\/tr>
    S&P 500 companies citing reputational risk from AI<\/td>\u2014<\/td>38%<\/td>Harvard Law School Forum on Corporate Governance[7]<\/sup><\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n

    Why AI Gets Pharma and Biotech Companies Wrong More Often Than Other Sectors<\/strong><\/h2>\n\n\n\n

    Training-data lag collides with a regulatory calendar<\/h3>\n\n\n\n

    A drug’s regulatory status, label, and even its dosing can change on a single FDA action date, but a language model’s knowledge of that change depends entirely on when it was trained and what got indexed afterward. Analysts covering AI visibility in pharma have documented that a drug approved or re-labeled in one year may not appear correctly in a chatbot’s answers until a year or more later, if it appears accurately at all.[19]<\/sup> That lag applies just as directly to an investor asking about pipeline status, trial results, or patent exclusivity as it does to a patient asking about dosing.<\/p>\n\n\n\n

    Even the FDA’s own AI tool hallucinates<\/h3>\n\n\n\n

    FDA employees testing the agency’s internal AI assistant, Elsa, told CNN it fabricates studies and cites regulatory sources that do not exist, describing the tool as confidently wrong on basic drug questions.[20]<\/sup> FDA Commissioner Marty Makary has publicly acknowledged the hallucination problem and confirmed Elsa’s use remains optional for staff.[21]<\/sup> This matters to investors in a specific way: if the regulator’s own internal AI cannot reliably summarize its own drug-approval record, an external chatbot synthesizing the same public record for an investor is working with a strictly harder problem and no internal quality control at all.<\/p>\n\n\n\n

    Mid-size and specialty biotech are nearly invisible to AI, which cuts both ways<\/h3>\n\n\n\n

    Vendor testing of AI responses about pharmaceutical brands has found that AI visibility skews heavily toward a small number of household-name drugs, while mid-size biotech companies \u2014 including ones with FDA-approved products treating large patient populations \u2014 are effectively absent from AI-generated answers.[19]<\/sup> For an investor, that produces two distinct failure modes rather than one: a large-cap name gets a confident, sometimes wrong answer built from thin or stale sourcing, while a smaller name gets no answer, an out-of-date answer, or a competitor’s product recommended instead. Testing of chatbot accuracy against authoritative drug-interaction references has separately found that systems can simultaneously miss real interactions and flag interactions that have no clinical basis, the same double-failure pattern.[19]<\/sup><\/p>\n\n\n\n

    What This Means for Investor Relations Teams<\/strong><\/h2>\n\n\n\n

    Investor relations has historically optimized for two channels: the company’s own IR website and the earnings-call transcript, both indexed for traditional search. AI-generated answers add a third channel the company does not control and, until recently, could not measure. Q4 Inc., which operates investor relations websites for more than half of the S&P 500 and supports over 2,600 public companies, launched a dedicated Answer Engine Optimization product for IR websites in March 2026, explicitly framing the goal as ensuring AI tools summarize a company’s results using its own IR materials rather than a secondary source.[22]<\/sup> Q4’s own guidance to IR teams reframes the operative question from “how do we rank for our ticker” to ensuring the company’s own site is the source an AI model draws on when summarizing quarterly results.[23]<\/sup><\/p>\n\n\n\n

    The gap between what ranks and what gets cited is already measured<\/h3>\n\n\n\n

    Ahrefs’ analysis of ChatGPT’s most-cited pages found that 28.3% of them have zero organic visibility in Google Search.[24]<\/sup> In other words, being well-optimized for traditional search does not guarantee an AI model will treat a company’s own materials as a source at all. For IR teams, that means the 10-K, the investor deck, and the press release archive can be technically excellent and still lose the AI-answer contest to a forum post, an analyst note, or a stale aggregator page.<\/p>\n\n\n\n

    What This Means for Medical Affairs, Regulatory Affairs, and Legal<\/strong><\/h2>\n\n\n\n

    An investor-facing AI error about a drug’s status is rarely a purely financial-communications problem. If an AI model tells an investor a trial met its primary endpoint when it did not, or that a drug carries an indication the FDA never approved, the same false claim is one a patient, a physician, or a plaintiff’s attorney could encounter using an identical prompt. That overlap is why DrugChatter’s AI-monitoring methodology \u2014 scheduled, multi-model querying of real questions against the current label and the current docket \u2014 was built around drug information rather than stock tickers, but produces exactly the underlying artifact an IR or legal team needs: a dated record of what a named AI model said about the company’s product, compared against the current authoritative source, at a specific point in time. Without that record, a company facing an investor complaint, a securities inquiry, or a physician’s question about the same misinformation has no way to show what was actually being said publicly, by which model, or since when.<\/p>\n\n\n\n

    A Practical Error Taxonomy for AI-About-Your-Company Monitoring<\/strong><\/h2>\n\n\n\n

    Distinguishing the kind of AI error at issue determines who inside the company should own the response.<\/p>\n\n\n\n

    Error type<\/th>Example<\/th>Primary owner<\/th><\/tr><\/thead>
    Stale regulatory status<\/td>AI describes a drug as investigational after FDA approval, or cites a superseded label<\/td>Regulatory Affairs \/ Medical Affairs<\/td><\/tr>
    Fabricated or misattributed trial result<\/td>AI states an endpoint was met, or attributes another company’s data to the wrong trial<\/td>Medical Affairs \/ Clinical<\/td><\/tr>
    Incorrect patent or exclusivity timeline<\/td>AI states a generic or biosimilar entry date that does not match the current Orange Book listing or docket<\/td>IP \/ Competitive Intelligence<\/td><\/tr>
    Fabricated legal or compliance history<\/td>AI states the company settled a regulatory action that did not occur<\/td>Legal \/ Investor Relations<\/td><\/tr>
    Missing visibility<\/td>AI has no answer, or defaults to describing a competitor instead<\/td>Investor Relations \/ Corporate Communications<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n

    Where AI and the Evidence Diverge<\/strong><\/h2>\n\n\n\n

    The pattern across every documented case in this piece is the same: the AI model states an answer with identical confidence whether the underlying source is a live FDA label, a two-year-old training snapshot, or nothing at all. Bard did not hedge its exoplanet claim.[9]<\/sup> ChatGPT did not flag its embezzlement summary as unverified.[10]<\/sup> Air Canada’s chatbot did not qualify its refund answer as possibly outdated.[14]<\/sup> That absence of calibrated uncertainty, more than the raw hallucination rate, is what turns an AI answer into an investor-relations exposure: a 5% error rate delivered with 100% confidence reads, to the person receiving it, exactly like a 0% error rate.<\/p>\n\n\n\n

    Key Takeaways<\/strong><\/h2>\n\n\n\n
      \n
    • 62% of retail investors already use AI tools in their investment research, and 53.5% specifically use chatbots such as ChatGPT.[1]<\/sup><\/li>\n\n\n\n
    • 72% of S&P 500 companies now disclose a material AI risk, up from 12% two years earlier, and 22% of Fortune 100 companies specifically name AI hallucination or misinformation as material.[6][7]<\/sup><\/li>\n\n\n\n
    • A single AI-generated factual error has already been tied to a documented ~$100 billion single-day market-value loss.[8]<\/sup><\/li>\n\n\n\n
    • Courts and tribunals have already held companies responsible for what their own deployed AI tools say, independent of intent.[14]<\/sup><\/li>\n\n\n\n
    • Pharma and biotech face a structurally worse version of this problem than most sectors, because regulatory status, trial results, and patent timelines change faster than model training cycles, and mid-size biotech names are frequently invisible to AI altogether.[19][20]<\/sup><\/li>\n\n\n\n
    • Answer Engine Optimization for investor relations is now a commercially available, actively adopted discipline, not a hypothetical one.[22]<\/sup><\/li>\n<\/ul>\n\n\n\n

      FAQ<\/strong><\/h2>\n\n\n\n

      Can an AI chatbot’s false statement about my company actually move the stock price?<\/h3>\n\n\n\n

      It already has. Alphabet lost roughly $100 billion in market value in a single trading day after its own Bard chatbot gave a factually wrong answer in a promotional demo, unrelated to Alphabet’s financials.[8]<\/sup> The mechanism was market perception of AI competitiveness, not a claim about Alphabet’s business directly, which means the exposure applies even when the false AI statement has nothing to do with the topic investors are actually worried about.<\/p>\n\n\n\n

      Is my company legally responsible for what a public AI chatbot like ChatGPT says about us?<\/h3>\n\n\n\n

      For a company’s own deployed chatbot, yes \u2014 the Air Canada ruling establishes that a company cannot disclaim liability for its own AI tool’s statements to customers.[14]<\/sup> For third-party AI models like ChatGPT or Gemini describing your company from public sources, liability runs to the AI provider under an evolving and still-unsettled defamation framework, but the reputational and investor-relations exposure is the company’s regardless of who is legally liable.[11][12]<\/sup><\/p>\n\n\n\n

      Do public companies have to disclose AI-related risks in their SEC filings?<\/h3>\n\n\n\n

      There is no standalone SEC rule requiring AI-specific disclosure as of September 2026, but the SEC’s existing materiality standard already applies, and the agency has brought enforcement actions against firms that overstated their AI capabilities.[15][17]<\/sup> Separately, 72% of S&P 500 companies now voluntarily disclose a material AI risk, effectively making it a market-standard disclosure even without a dedicated rule.[7]<\/sup><\/p>\n\n\n\n

      Why do AI chatbots get pharmaceutical companies wrong more often than other industries?<\/h3>\n\n\n\n

      Drug regulatory status, label language, trial results, and patent exclusivity all change on specific, dated events that outrun a language model’s training cutoff, and independent audits already show FDA’s own internal AI tool hallucinating fabricated studies.[19][20]<\/sup> Mid-size and specialty biotech companies also have thinner public-source coverage than large consumer brands, which independent testing shows correlates with higher AI visibility gaps and error rates.[19]<\/sup><\/p>\n\n\n\n

      What is Answer Engine Optimization (AEO) for investor relations?<\/h3>\n\n\n\n

      AEO is the practice of structuring an IR website and disclosure materials so that AI models cite the company’s own primary sources when summarizing its results, rather than a secondary or outdated source. Q4 Inc., which runs IR websites for more than half of the S&P 500, launched a dedicated AEO product for IR sites in March 2026.[22]<\/sup><\/p>\n\n\n\n

      Can being well-optimized for Google search guarantee accurate AI answers about my company?<\/h3>\n\n\n\n

      No. Ahrefs found that 28.3% of ChatGPT’s most-cited pages have no organic visibility in Google Search at all, meaning traditional SEO performance and AI-citation performance are measurably different problems requiring separate monitoring.[24]<\/sup><\/p>\n\n\n\n

      What is the SEC’s “AI washing” enforcement focus, and does it apply to biopharma issuers?<\/h3>\n\n\n\n

      AI washing refers to companies overstating their actual AI capabilities in public statements. The SEC’s first enforcement actions targeted investment advisers Delphia and Global Predictions in March 2024 for a combined $400,000 in penalties.[17]<\/sup> The same reasonable-basis standard applies to any issuer, including biopharma companies, describing their own AI use in drug discovery, diagnostics, or operations.[18]<\/sup><\/p>\n\n\n\n

      Has anyone successfully sued an AI company for defamation over false business claims yet?<\/h3>\n\n\n\n

      Not conclusively. In Walters v. OpenAI, the plaintiff lost because he could not meet the fault standard required of a public figure and because the recipient of the false statement recognized it as unreliable.[11]<\/sup> Wolf River Electric’s suit against Google over an allegedly false AI Overview claim about a settled deceptive-practices lawsuit remains an open, closely watched case as a business (not individual) plaintiff alleging concrete financial harm.[13]<\/sup><\/p>\n\n\n\n

      How is monitoring AI answers about drug information different from monitoring AI answers for investors?<\/h3>\n\n\n\n

      The audience and stakes differ, but the underlying error taxonomy is largely the same: stale regulatory status, fabricated or misattributed trial results, and incorrect timelines. A monitoring approach built around scheduled, multi-model querying against the current label and docket produces evidence useful to Medical Affairs, Regulatory Affairs, Legal, and Investor Relations simultaneously, because all four functions are exposed by the same underlying AI error.<\/p>\n\n\n\n

      What should an IR team do first if it suspects AI models are describing the company inaccurately?<\/h3>\n\n\n\n

      Establish a dated baseline: query the major consumer AI platforms with the specific questions investors are likely to ask (pipeline status, trial results, patent timeline, litigation history), record the exact answers and which model produced them, and compare each claim against the current 8-K, label, or docket entry. That record is what turns “we think AI is saying something wrong about us” into a specific, actionable, and defensible finding.<\/p>\n\n\n\n

      References<\/h2>\n\n\n\n
        \n
      1. Investing.com. “How Retail Investors Are Using AI in 2026.” Investing.com Blog, April 5, 2026. https:\/\/www.investing.com\/blog\/how-retail-investors-are-using-ai-in-2026-339<\/li>\n\n\n\n
      2. Vectara Hallucination Index, cited in: Metricus. “AI Visibility for Pharma: When AI Chatbots Get Drug Information Wrong.” May 17, 2026. https:\/\/metricusapp.com\/blog\/pharmaceutical-ai-visibility\/<\/li>\n\n\n\n
      3. Claritywithai.org. “Best AI Tools for Stock Market Investors in 2026.” July 11, 2026. https:\/\/www.claritywithai.org\/2026\/06\/202606best-ai-tools-stock-market-investors-2026.html<\/li>\n\n\n\n
      4. Modern Finance. “ChatGPT, financial literacy, and individual investment decisions.” Vol. 4, No. 2, 2026. https:\/\/mf-journal.com\/article\/view\/539<\/li>\n\n\n\n
      5. The AI Corner. “How to Use AI for Stock Investing: Tools and Prompts 2026.” April 1, 2026. https:\/\/www.the-ai-corner.com\/p\/how-to-use-ai-for-stock-investing-tools-prompts-2026<\/li>\n\n\n\n
      6. Henderson, Lee, and Jamie Smith (EY Center for Board Matters). “How Boards Can Lead in a World Remade by AI.” Harvard Law School Forum on Corporate Governance, February 19, 2026. https:\/\/corpgov.law.harvard.edu\/2026\/02\/19\/how-boards-can-lead-in-a-world-remade-by-ai\/<\/li>\n\n\n\n
      7. “AI Risk Disclosures in the S&P 500: Reputation, Cybersecurity, and Regulation.” Harvard Law School Forum on Corporate Governance, October 15, 2025. https:\/\/corpgov.law.harvard.edu\/2025\/10\/15\/ai-risk-disclosures-in-the-sp-500-reputation-cybersecurity-and-regulation\/<\/li>\n\n\n\n
      8. NPR. “Google shares drop $100 billion after its new AI chatbot makes a mistake.” February 9, 2023. https:\/\/www.npr.org\/2023\/02\/09\/1155650909\/google-chatbot-error-bard-shares<\/li>\n\n\n\n
      9. KTVZ \/ CNN (Catherine Thorbecke). “Google’s AI chatbot Bard had an inaccurate response in public demo.” February 8, 2023. https:\/\/ktvz.com\/lifestyle\/technology\/cnn-social-media-technology\/2023\/02\/08\/googles-ai-chatbot-bard-had-an-inaccurate-response-in-public-demo\/<\/li>\n\n\n\n
      10. Columbia Undergraduate Law Review. “Redefining Defamation: Establishing Proof of Fault for Libel and Slander in AI Hallucinations.” December 14, 2025. https:\/\/www.culawreview.org\/journal\/redefining-defamation-establishing-proof-of-fault-for-libel-and-slander-in-ai-hallucinations<\/li>\n\n\n\n
      11. Terms.law. “Hallucinations vs Defamation: When AI Just Makes Stuff Up About You.” December 6, 2025. https:\/\/terms.law\/2025\/12\/05\/hallucinations-vs-defamation-when-ai-just-makes-stuff-up-about-you\/<\/li>\n\n\n\n
      12. Quinn Emanuel. “Client Alert: Defamation in the AI Era.” March 9, 2026. https:\/\/www.quinnemanuel.com\/the-firm\/publications\/client-alert-defamation-in-the-ai-era\/<\/li>\n\n\n\n
      13. Quinn Emanuel. “Client Alert: Defamation in the AI Era” (Wolf River Electric v. Google LLC discussion). March 9, 2026. https:\/\/www.quinnemanuel.com\/the-firm\/publications\/client-alert-defamation-in-the-ai-era\/<\/li>\n\n\n\n
      14. Lex (Substack). “AI: Who is responsible if your AI hallucinates financial mistakes?” https:\/\/lex.substack.com\/p\/ai-who-is-responsible-if-your-ai<\/li>\n\n\n\n
      15. Finrep. “SEC AI Disclosure Requirements for the 10-K: 2026 Practitioner Guide.” July 20, 2026. https:\/\/www.finrep.ai\/blog\/sec-ai-disclosure-requirements-for-the-10k-2026-practitioner-guide<\/li>\n\n\n\n
      16. Context Analytics. “The Quiet Part: What 10-K Risk Factors Say About AI in 2026.” July 8, 2026. https:\/\/www.contextanalytics-ai.com\/case-studies\/the-quiet-part-what-10-k-risk-factors-say-about-ai-in-2026\/<\/li>\n\n\n\n
      17. Latham & Watkins. “SEC Announces First-Ever Enforcement Actions for ‘AI Washing.'” March 25, 2024. https:\/\/www.lw.com (client alert PDF)<\/li>\n\n\n\n
      18. Verdict. “US regulator charges investment advisers over AI washing.” 2024. https:\/\/www.verdict.co.uk\/us-regulator-charges-investment-advisers-over-ai-washing\/<\/li>\n\n\n\n
      19. Metricus. “AI Visibility for Pharma: When AI Chatbots Get Drug Information Wrong.” May 17, 2026. https:\/\/metricusapp.com\/blog\/pharmaceutical-ai-visibility\/<\/li>\n\n\n\n
      20. Vice. “The FDA’s AI Is Busy Approving Drugs\u2014and Hallucinating Fake Studies.” July 25, 2025. https:\/\/www.vice.com\/en\/article\/the-fdas-ai-is-busy-approving-drugs-and-hallucinating-fake-studies\/<\/li>\n\n\n\n
      21. Applied Clinical Trials Online. “FDA’s Elsa AI Tool Raises Accuracy and Oversight Concerns.” 2026. https:\/\/www.appliedclinicaltrialsonline.com\/view\/fda-elsa-ai-tool-raises-accuracy-and-oversight-concerns<\/li>\n\n\n\n
      22. Business Wire. “Q4 Launches AEO for IR Web to Help Public Companies Stand Out in AI-Generated Answers.” March 3, 2026. https:\/\/www.businesswire.com\/news\/home\/20260303509243\/en\/Q4-Launches-AEO-for-IR-Web-to-Help-Public-Companies-Stand-Out-in-AI-Generated-Answers<\/li>\n\n\n\n
      23. Q4 Inc. “The Future of IR Websites: Optimizing for AI-Driven Discovery.” July 18, 2026. https:\/\/www.q4inc.com\/resource-center\/blog\/the-future-of-ir-websites-optimizing-for-ai-driven-discovery<\/li>\n\n\n\n
      24. Erlin.ai. “Generative Engine Optimization Trends for 2026.” May 15, 2026. https:\/\/www.erlin.ai\/blog\/generative-engine-optimization-trends<\/li>\n<\/ol>\n","protected":false},"excerpt":{"rendered":"

        An investor researching a biopharmaceutical company in 2026 is more likely to open ChatGPT, Perplexity, or Gemini than to read […]<\/p>\n","protected":false},"author":1,"featured_media":1250,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_lmt_disableupdate":"","_lmt_disable":"","site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[1],"tags":[],"class_list":["post-1249","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-general"],"modified_by":"DrugChatter","_links":{"self":[{"href":"https:\/\/drugchatter.com\/insights\/wp-json\/wp\/v2\/posts\/1249","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/drugchatter.com\/insights\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/drugchatter.com\/insights\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/drugchatter.com\/insights\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/drugchatter.com\/insights\/wp-json\/wp\/v2\/comments?post=1249"}],"version-history":[{"count":1,"href":"https:\/\/drugchatter.com\/insights\/wp-json\/wp\/v2\/posts\/1249\/revisions"}],"predecessor-version":[{"id":1251,"href":"https:\/\/drugchatter.com\/insights\/wp-json\/wp\/v2\/posts\/1249\/revisions\/1251"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/drugchatter.com\/insights\/wp-json\/wp\/v2\/media\/1250"}],"wp:attachment":[{"href":"https:\/\/drugchatter.com\/insights\/wp-json\/wp\/v2\/media?parent=1249"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/drugchatter.com\/insights\/wp-json\/wp\/v2\/categories?post=1249"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/drugchatter.com\/insights\/wp-json\/wp\/v2\/tags?post=1249"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}