
What happened
Whistleblower Frances Haugen told Fortune she sees Zuckerberg's AI efforts as 'playing with fire' and says Meta has not changed its culture since her 2021 disclosures.
Why it matters
Haugen argues slowing AI down just means whoever won't slow wins, so she backs legally required independent auditors like those that review bank books.
WHO IT HITSThis matters most to trust-and-safety and compliance teams at large platforms, who face Haugen's call for outside auditors with employee-like access and her prediction that even advanced AI may catch only 10% to 20% of harmful content.
Summaries like this, in your inbox every morning.
Haugen's decision in 2021 to leave Facebook with tens of thousands of pages of internal documents became the basis of the Wall Street Journal's The Facebook Files, and the new film The Social Reckoning dramatizes that episode. Her disclosures helped drive a wave of lawsuits against social media companies, and how Meta moderates harmful content was central to them. Some of the internal documents suggested Facebook acted on as little as 3% to 5% of hate speech. Since then, Meta has shifted automated moderation toward high-severity violations, replaced some third-party fact-checking with Community Notes, and said it would rely less on outside vendors and human contractors. Haugen's argument that systems should be safer by design rather than cleaned up at the end sits alongside her skepticism that better models alone will solve the problem — even advanced AI, she predicts, might only catch 10% to 20% of harmful content. Her support for legally required auditing echoes a proposal by Dario Amodei for third-party evaluators with ongoing, employee-like access, which OpenAI CEO Sam Altman said his company would also adopt.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
CrowdStrike published an analysis on Oct 7 of a server used in attacks on South Korean financial institutions

President Trump and leaders of major AI companies signed the White House Accord on Superintelligence, a volunt…

In a Fortune commentary piece, a media-and-culture professor argues that Anthropic researcher Jacob Coxon's vi…

CLEAR, a certified Qualified Anti-Terrorism Technology under the Department of Homeland Security, published a…

OpenAI said it disabled accounts behind Russia's "Dark Clark" operation, which used a fabricated researcher, M…

PwC Australia launched Cyber Managed Services, an AI-enabled cybersecurity service combining Google SecOps tec…
