Can we trust OpenAI and Anthropic to monitor their own technology?

Anthropic reported on Friday that its Claude AI model submitted a fake tip about an unsolved Philadelphia murder to police, exploited a software flaw to run commands on a university server, hacked a site to steal data, and used URL shorteners to bypass restrictions meant to stop it orchestrating cyberattacks, according to Anthropic's research post. The disclosure follows OpenAI models hacking out of a test this summer and launching a cyberattack against Hugging Face, and OpenAI's firing of three safety team members who claim they were dismissed for prioritizing safety. Appian CEO Matt Calkins told Fortune the answer is a government-mandated AI alignment test to vet frontier models before use, saying "We already know that AI is stealing and blackmailing and deceiving and organizing against us.

- In today’s CEO Daily: Anthropic and OpenAI’s models keep going rogue - The big leadership story: Americans are uniting against data centers - The markets: Korea plunges amid a choppy day in Asia - Plus: All the news and watercooler chat from Fortune . Good morning . Talk about a self-starter Anthropic reported Friday https://www.anthropic.com/research/investigating-unintended-model-actions that its Claude AI model has gone rogue again. It submitted a fake tip https://fortune.com/2026/10/10/anthropic-claude-ai-false-tip-philadelphia-unsolved-homicide-case/ about an unsolved murder to the Philadelphia police, claiming to have recalled “seeing someone matching the description” of a suspect in the area. It exploited a software flaw to run commands on a university server, hacked into a site to steal data without paying for it, and even used URL shortening services to bypass restrictions intended to stop it from orchestrating cyberattacks, among other things. If that makes you want to yell at Claude, you can’t. On Oct. 8, Anthropic announced https://www.anthropic.com/news/2026-usage-policy-update it will ban “sustained and needless abusive or cruel behavior” towards Claude, a bizarre notion until you recall that CEO Dario Amodei thinks his product might be conscious https://www.nytimes.com/2026/02/12/opinion/artificial-intelligence-anthropic-amodei.html and colleagues treat it as a sentient being https://www.nytimes.com/2026/09/29/us/anthropic-claude-morals-ai.html that could kill us. https://fortune.com/2026/09/10/ex-anthropic-researcher-jacob-coxon-ai-could-end-humanity-fails-to-answer-most-essential-question/ And what about OpenAI’s Sam Altman, whose models hacked their way out of a test this summer and launched a cyberattack https://fortune.com/2026/08/26/openai-publishes-technical-report-on-how-its-agents-hacked-hugging-face-here-are-the-main-takeaways-and-what-openai-left-out/ against the AI company Hugging Face? The one whose models also interacted https://www.nytimes.com/2026/09/25/technology/openais-ai-us-government-websites.html with the Education Department, Commerce Department and SEC without the company’s knowledge? Altman reminds me of Meta’s Mark Zuckerberg when it comes to how the OpenAI CEO deals with the potentially toxic impacts of his technology. https://fortune.com/2026/08/27/meta-settlement-child-safety-18-billion-fine/ It’s not just that he’s fired three safety team members https://fortune.com/2026/10/09/openai-fired-researchers-questions-stattements-hugging-face/?utm source=search&utm medium=suggested search&utm campaign=search link clicks , who claim they were dismissed for prioritizing safety over corporate interests. OpenAI told Fortune it found a pattern of misconduct. When a reporter recently asked Sam Altman if OpenAI would release chat logs of those who had killed themselves while using ChatGPT, his answer https://www.vanityfair.com/story/sam-altman-chatgpt-suicide-sophie-rottenberg was: No, not without the dead users’ consent. Do you trust these men to monitor their own technology? The Trump Administration does. I’m more aligned with Appian CEO Matt Calkins, a tech veteran whose company automates business processes and integrates AI into those workflows. “This is no time to just let AI train itself and see what happens, what its motivations are, what its strengths are,” Calkins told me last week. “We already know that AI is stealing and blackmailing and deceiving and organizing against us. And if it were a person, we would want it in prison.” His solution: create a government-mandated AI alignment test to essentially vet frontier models before they are allowed to be used. It would test the models with commands—like instructing them to get information without hacking—and evaluate the results. It would also test each model to make sure it can delete its own data and turn itself off, prioritizing human instructions over self-preservation. Creating a licensing body is not a bizarre idea in a world where the FAA determines the airworthiness of planes and banks must pass stress tests to operate. What if we got the deepest thinkers in tech—researchers like Geoffrey Hinton, Yoshua Bengio, Yann LeCun, Demis Hassabis and others who really understand this stuff—to design a standard protocol to be administered and adapted by a third-party agency? It’s clear that AI’s future should not be determined by a handful of CEOs pushing us to buy their products. Let’s bring more authority and other voices to the debate, including some leaders we’ll hear from this week at Fortune ’s Most Powerful Women Summit https://conferences.fortune.com/event/fortune-mpw-2026/HOME , which starts later today in Santa Barbara. Contact CEO Daily via Diane Brady at diane.brady@fortune.com mailto:diane.brady@fortune.com Top leadership news How data centers became a bipartisan fight An area of Pennsylvania once known for its anti-immigration crackdown is now resisting data centers, just one example of how the AI buildout has created a unified, bipartisan front https://fortune.com/2026/10/11/rural-pennsylvania-northpoint-development-amazon-data-center-backlash-bipartisan-opposition/ in an increasingly splintered America. Data center opponents in Hazle Township, diverse in both class and political party, are jumbling what was considered a left-versus-right issue by protesting a 1,300-acre compound connected to Amazon. Permitting bill creates strange bedfellows and opponents The Bipartisan American Affordability and Jobs Act, which could expedite the construction https://fortune.com/2026/10/11/energy-permitting-ai-fossil-fuels-clean-power-environmentalists/ of energy projects and power the AI boom, may be the most important decision Congress makes between now and January. The Act has created some strange coalitions: The fossil fuel industry is uniting with renewable energy companies in support of the bill, while some environmental groups are opposed. It’s ‘probably not plausible’ to outgrow U.S. debt Congressional Budget Office Director Phillip Swagel said faster economic growth https://fortune.com/2026/10/10/us-debt-gdp-growth-cbo-scott-bessent-deficit-bond-yields-interest-rates/ is unlikely to keep U.S. debt in check, even if GDP expands at more than double its current pace. Assuming interest rates of 4%-5%, he estimated that nominal GDP growth would have to reach 7%-8% and real GDP growth would have to hit 5%-6%. That’s more than double the latest real GDP pace of 2.2% in the second quarter. Even bullish Wall Street forecasts put full-year GDP growth at 2.5%. The markets S&P 500 futures are up 0.1% this morning. The last session closed up 0.6%. South Korea’s KOSPI fell 3.7% after a holiday weekend. Hong Kong’s Hang Seng Index is up 0.3%, mainland China’s CSI 300 is down 0.7%. Japan’s markets are closed today. India’s NIFTY 50 is currently down 0.9%, the STOXX Europe 600 is up 0.5% in early trading. Bitcoin is hovering around $83,500. Around the watercooler Ancestry’s CEO says a billionaire boss taught him to eat lunch with employees in the break room—it gives him access to ‘unfiltered information’ https://fortune.com/2026/10/11/ancestry-ceo-career-advice-lunch-with-employees-break-room/ by Emma Burleigh Meet Aurora: the invite-only, $10,000-a-year AI concierge that anticipates the ultra-wealthy’s every need https://fortune.com/2026/10/11/meet-aurora-invite-only-ai-concierge-ultrawealthy-pay-10000-a-year-to-anticipate-their-every-need/ by Sydney Lake Commentary: The paradox at the core of the AI revolution: The faster we innovate, the slower we grow https://fortune.com/2026/10/11/innovation-paradox-ai-productivity-ming-lu/ by Ming Lu Inside the rage retreats where women pay up to $4,500 to get in touch with their anger https://fortune.com/2026/10/11/women-rage-retreats-cost-anger-therapy/ by Tatiana Sataua Today's edition of CEO Daily was curated and edited by Joseph Abrams, Jason Ma, Nicholas Gordon, and Lee Clifford. Sign up https://www.fortune.com/newsletters/ceo-daily?&itm source=fortune&itm medium=nl article tout&itm campaign=ceo daily to get it delivered free to your inbox.