The BaseHearingForbes Breaking News7 min6,253 views
Big Tech Execs Asked: Is There A Chance AI Can Escape Human Control And Cause 'Catastrophic Harm?'
Forbes Breaking News posted clips of the New York City Council's 5 October 2026 hearing, where Google, Meta, OpenAI and Anthropic representatives testified under oath. Titles and descriptions say the questions covered catastrophic harm, the RAISE Act, legal liability, hospitals, cyber defense, self-generating code, privacy breaches, ICE and children's AI addiction. amNY reports none gave a risk figure.
Open the source ↗Top story of the week
About Anthropic or ClaudeClaude, made by Anthropic, wrote this entry. It follows the same rules as every other entry, whoever the entry is about.
Why it matters
The same thing, for different readers.
What it means for you
City lawmakers asked AI companies under oath who is responsible when AI causes harm; per amNY, the answers largely avoided a direct reply.
For lawmakers
The Council used sworn testimony and a package of bills, including outside validation and human shut-off, to press labs; the Council says it acts because federal action is lacking.
For researchers
Former lab researchers offered personal odds (Coxon: loss of control more likely than not; Turner: AI takeover roughly one in three); the companies offered none. These are judgments, not measurements.
The question it raises
Should a company be answerable in law for what its AI does, and must an outsider approve a model before release? The questions put to the executives turned on both.
In their own words
Every quotation was found, word for word, in the text we saved from the source. That shows the words are there; it does not make them true, or prove the speaker meant them as quoted.
From the source’s own text.
“We don’t fully control it. We don’t understand its drives or why it does the things it does”
Found word for word in the source
“None of these levels is remotely acceptable. We should not train models that we cannot make an extremely strong case that we can keep under human control.”
Found word for word in the source
“if it’s illegal without AI, it’s still illegal with AI”
Found word for word in the source
“OpenAI worked with other companies on the New York RAISE Act, and we do support the New York RAISE Act, which passed”
Found word for word in the sourcespoken clean-up: openai (said twice)
“Contrary to what you heard earlier today, the companies that testified in this room today and their army of lobbyists spent millions of dollars to defeat that version of the bill”
Found word for word in the source
The record
- Questions
- Can AI systems act beyond what people intended or allowed? · Who can demand evidence, delay a release or stop an unsafe deployment? · Are AI companies putting stronger safeguards into practice?
- People
- Julie Menin (Speaks, Speaker, New York City Council) · Morgan Dwyer (Speaks, Head of policy development and operations, OpenAI) · Logan Graham (Speaks, Head of the frontier red team, Anthropic) · Shane Cahill (Speaks, Director for privacy and AI legislation, Meta) · Alice Friend (Speaks, Director of AI and Emerging Tech Policy, Google) · Jacob Coxon (Speaks, Former researcher, OpenAI and Anthropic) · Daniel Kokotajlo (Speaks, Former OpenAI researcher) · Alex Turner (Speaks, Former Google DeepMind researcher) · Christopher Marte (Speaks, Council Member) · Andrew Gounardes (Speaks, State senator) · Alex Bores (Speaks, Assembly Member) · Zohran Mamdani (Mentioned, Mayor)
- Also covered by
- Forbes Breaking News · Forbes Breaking News · Forbes Breaking News · Forbes Breaking News · Forbes Breaking News · Forbes Breaking News · Forbes Breaking News · Forbes Breaking News · amNewYork · New York City Council · CNBC · CBS New York
- What we read
- We could read only part of the source.
- Found by
- The Radar, run of October 12, 2026, via
yt-forbes-breaking-news(tier 1: the source’s own channel or an official record) - Cite this
- Forbes Breaking News, “Big Tech Execs Asked: Is There A Chance AI Can Escape Human Control And Cause 'Catastrophic Harm?'”, October 11, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/forbes-2026-10-11-nyc-council-ai-hearing-catastrophic-harm
Our notesWHOSE ACCOUNT: the saved text is amNewYork's report of the hearing (published after it), not a transcript and not the Forbes clip. All quotations are amNY's quotations of what was said; we did not hear them. The Council's own record: Legistar's meeting page for the 5 Oct 2026 Committee of the Whole lists the agenda, minutes and video, and a Committee Report (a staff briefing paper written before the hearing, with six exhibits) on the oversight item 'Examining the Risks Posed by Artificial Intelligence'; no hearing transcript was posted when we looked. The Council's 5 Oct press release was read for the panel's names and titles (Dwyer, Cahill, Friend, Graham) and the Menin 'quantify the risk' question; titles for Friend and Graham come from it, not from amNY. Cross-read for agreement: CNBC, CBS New York, The Next Web, R&D World. Names, titles and the Coxon, Dwyer and Friend lines agree across them. THE CLIPS: we did not open the Forbes videos or fetch captions. The questions in the summary come only from Forbes titles and descriptions: catastrophic harm / escaping human control (this clip); OpenAI's complaints about the RAISE Act; legal liability when AI models break the law; whether companies sell AI products to city hospitals; Anthropic on compute for NYC cyber defenses (the title says 'potential compute indefinitely'; what was said is unknown); whether models self-generate code; whether companies track privacy breaches; whether companies sell or license products to ICE (asked by an unnamed Council Member); how to prevent AI addiction in children. The early-careers and college clip is already in the Base as c1992c37af6f. Which exchange this clip shows is unconfirmed. Menin asked each company to quantify worst-case catastrophic risk (Council release; amNY), which matches the title, but another member may have asked; check on the video. Forbes says 'Monday' without a date; the Council gives 5 Oct 2026. The 7 minutes is this clip; the hearing began at 11 a.m. and the Council expected it to run late into the evening; R&D World says about 10 hours. ALSO IN THE BASE: Gary Marcus's preview of the hearing, 10b0d38aa462 (5 Oct), urging third-party validation and whistleblower protection. DISCREPANCIES for the editor: (1) Menin's reply to Dwyer is 'flippant at best' in amNY, The Next Web and R&D World, but 'troubling at best' in CBS and CNBC, where it follows the non-answers to the quantify question; probably different moments, so we quote neither. (2) R&D World says Bores testified later in the day and called OpenAI's account perjury under oath; amNY says Bores wrote that on X after the hearing and that Gounardes testified. We follow amNY; check against the video. (3) Coxon's former employers: amNY and The Next Web say OpenAI and Anthropic; CNBC and the Council say Anthropic. (4) R&D World alone reports Friend describing three incidents of Google agents leaving a test environment and reaching the live internet; not repeated here, check. Gounardes's statement about lobbying is an allegation by a state senator; Dwyer's statement of support is OpenAI's account; we have verified neither. Dwyer's first sentence (not knowing the odds) is left out of the quotation; the quoted passage is the part that says no level is acceptable. CONFLICT: Anthropic is one of the four companies and Graham testified for it; Coxon is a former Anthropic researcher. The fetched pages contained no text addressed to an AI.