Whistleblower warns that humans don’t control AI

Read the original at Al Jazeera ↗
Al Jazeera · collected 2026-10-05

Quick Summary

Jacob Coxon, a former researcher at Anthropic, has warned during testimony before the New York City Council that artificial intelligence companies do not fully control their models and are proceeding too rapidly without sufficient safeguards. Coxon asserts that upcoming AI systems will likely be capable of enhancing themselves independently, rendering human oversight increasingly obsolete.
Written locally by qwen2.5:14b on 2026-10-06, using this article's own text rather than the other coverage of the same event (that is the story summary below).

AI analysis runs on qwen2.5:14b, locally

Story summary

Jacob Coxon, a 27-year-old former researcher at Anthropic and OpenAI, testified before the New York City Council on Monday alongside safety representatives from tech giants like Google, Meta, and OpenAI. Coxon, who quit his job over safety concerns last September, warned about the risks of AI companies racing to develop technology without proper safeguards in place. He stated that these firms do not fully control their AI models and are moving too quickly towards creating systems capable of self-improvement, which could pose a significant threat to humanity. The hearing, requested by Council Speaker Julie Menin, aimed to address potential legislation for establishing AI safety measures as city officials consider bills to regulate the industry. Coxon's testimony highlighted concerns over potential catastrophic scenarios, emphasizing that companies must better quantify and manage risks associated with advanced AI technologies.

Written for “AI Safety Testimony NYC” on 2026-10-06, grounded in this article and the 4 other(s) covering the same event.
Why this leaning score
The model judged this article politically coded and scored it -0.30, but all 1 of its quote(s) are attributed speech - words the article quotes from someone, not the article's own narration, so the score is not published.
Written under an earlier scoring contract, which gave a paragraph rather than checkable quotes. Re-analysing this article replaces it.
Leaning score withheld for article 58803: attributed speech only · logged 2026-10-06

Signals How these are calculated →

Claims extracted
3
claim-shaped sentences
Uncertain
0%
0 of 3 hedged
Leaning
withheld
no quote in the article backed the model's score
Correction & hedging signals
60.3
corrections and hedging in what we collected; not a measure of accuracy
Outlets on this story
5
Technology
Narrative spread
1
articles carrying this framing
Analyzed 2026-10-06 · how these are computed

Story

📰 AI Safety Testimony NYC
Technology · 5 article(s) covering the same event. See how they differ ↓

How this is being covered How these are calculated →

Article leaning vs. publisher reliability
Source leaning vs. consistency

Compared with similar articles

This article reads unscored and hedges 0% of its claims. Each row says how that neighbour differs.
We Won't Know the Answers to AI's Most Important Questions Until It's Too Late different event · 95%
TIME
⚖️ Leans strongly left 🔴 24% hedged 24 of 100 📰 publisher trust 60
“The articles discuss different perspectives on AI risks and development timelines but do not describe the same specific incident or occurrence.”
Former Anthropic security leader warns AI agents are becoming too autonomous for humans to keep them in check different event · 95%
Fox News
⚖️ leaning not scored 🔴 22% hedged 8 of 36 📰 publisher trust 69
“The articles describe different individuals (Jeffrey Ladish and Jacob Coxon) expressing concerns about AI control at distinct times and contexts.”
Ex-Anthropic employee who quit over safety concerns to testify at NYC council hearing same event · 95%
Washington Examiner
⚖️ Leans left 🔴 24% hedged 4 of 17 📰 publisher trust 72
“Both articles describe Jacob Coxon testifying at a New York City council hearing about AI safety concerns on the same day.”
AI industry insiders voice alarms to NYC council as local governments take up safety concerns same event · 95%
Toronto Star
⚖️ Leans left 🔴 50% hedged 1 of 2 📰 publisher trust 63
“Both articles describe testimony given by a whistleblower at a New York City Council hearing on Monday about AI safety concerns.”
Do AI and air traffic control mix? CEO addresses anxiety about new FAA tool different event · 90%
NPR
⚖️ leaning not scored 🔴 0% hedged 0 of 43 📰 publisher trust 60
“The articles discuss different aspects of AI concerns but describe distinct events and times.”
AI agents are "very intelligent, charming sociopaths," Kevin Roose says different event · 90%
CBS News
⚖️ leaning not scored 🔴 12% hedged 7 of 60 📰 publisher trust 66
“The articles discuss different aspects of AI control and development over time rather than a single specific incident.”
Trump says AI can self-regulate. Even some top Republicans don’t trust that different event · 90%
The Independent
⚖️ leaning not scored 🔴 14% hedged 5 of 35 📰 publisher trust 59
“The articles discuss different aspects of AI regulation and control over separate dates.”
Big Tech is leading the ‘AI safety’ guardrails push: For protection or profit? different event · 80%
Washington Examiner
⚖️ Leans left 🔴 21% hedged 12 of 58 📰 publisher trust 72
“The articles discuss different aspects of warnings about AI safety and control, but they do not describe the same specific incident occurring at the same time and place.”
Activist network wants you scared of AI killing people — unless China’s using it different event · 75%
Washington Examiner
⚖️ leaning not scored 🔴 23% hedged 10 of 43 📰 publisher trust 72
“The articles describe different events involving Jacob Coxon's warnings about AI but at distinct times and places with different outcomes.”
They’re profiting off your panic: How AI fear became a business model different event · 70%
Washington Examiner
⚖️ Leans left 🔴 6% hedged 3 of 47 📰 publisher trust 72
“The articles describe related events but not the exact same incident. Article A discusses a warning thread that gained significant attention and its aftermath, while Article B focuses on a whistleblower's testimony before the New York City council.”

Publisher

Al Jazeera · 2203 article(s) · 2 correction(s) detected
Running correction rate · 2 correction(s)
2026-10-03
Tennessee prisons official resigns after Christa Pike’s failed US execution
2026-09-30
Indonesia suspends five officials over luxury apartments for inmates

Who wrote this

No reporter is named on this article.

Topics

Anthropic New York City

Subjects

Anthropic ORG · 1× Coxon PERSON · 1× Jacob Coxon PERSON · 1× New York City GPE · 1×

Narrative

Speaking to the New York City council, Coxon says AI systems will soon be able to improve themselves, making human research redundant.
framing: assertive · carried by 1 article(s) · first seen 2026-10-06
🔮 Speaking to the New York City council, Coxon says AI systems will soon be able to improve themselves, making human research redundant.
2026-10-06 · Al Jazeera
Whistleblower warns that humans don’t control AI · assertive framing

Claims (3 extracted, 0 hedged)

Whistleblower warns that humans don't control AI asserted
humans → warn → AI
Former Anthropic researcher Jacob Coxon warns that AI companies ‘don’t fully control’ their models and are pushing ahead too quickly. asserted
companies → warn → models
Speaking to the New York City council, Coxon says AI systems will soon be able to improve themselves, making human research redundant. asserted
research → speak → themselves
💬Give feedback
🕘History 🎫Support
Terms · Privacy · beta v2.0.0
Articles to 6 Oct 2026