Photo via NPR

OpenAI announces new AI safety program; researchers worry it won't be enough

Read the original at NPR ↗
Cite

Jingnan, H. (2026, November 18). OpenAI announces new AI safety program; researchers worry it won't be enough. NPR. https://www.npr.org/2026/11/18/nx-s1-6044223/openai-embedded-evaluators-program-critics

Dated from the outlet's own feed. Check the original for its byline.

NPR · collected 2026-11-18 · by Huo Jingnan

Quick Summary

OpenAI has announced a new AI safety program that will involve embedded assessors from four AI safety and research organizations. However, AI safety researchers are skeptical, arguing that the program falls short of CEO Sam Altman’s previous promise of giving independent evaluators "employee-like access" to internal systems. The new program includes limited scopes for assessors and restricted access to employees and internal communication channels, leading to doubts about its effectiveness in addressing potential risks associated with advanced AI systems.
Written by AI on 2026-11-18, from this article's own text only (the story summary below draws on all the coverage).
See how they differ ↓

AI analysis

Story summary

OpenAI announced on November 17, 2026, that four independent assessors from research organizations like METR, Apollo Research, Transluce, and Goodfire would evaluate its internal AI models and safety features. However, critics argue that the assessors do not have full access to OpenAI’s technology and that the company retains control over which staff members the evaluators can interview. This announcement comes after months of security issues and declining trust in OpenAI's AI technology, including a major security incident at Hugging Face in July. AI researchers and experts are concerned that the new program falls short of CEO Sam Altman’s September promise of providing independent evaluators with "employee-like access" to internal systems.

Written for “OpenAI's AI Safety Program Criticized” on 2026-11-21, grounded in this article and the 1 other(s) covering the same event.

How they differ

2 outlets covered “OpenAI's AI Safety Program Criticized”.

What each outlet says

One AI summary per outlet of its latest article on this story, written from that article alone and checked against it. Open the article for the outlet's own words.

NPRThis outlet18 Nov

OpenAI has announced a new AI safety program that will involve embedded assessors from four AI safety and research organizations. However, AI safety researchers are skeptical, arguing that the program falls short of CEO Sam Altman’s previous promise of giving independent evaluators "employee-like access" to internal systems. The new program includes limited scopes for assessors and restricted access to employees and…

OpenAI announces new AI safety program; researchers worry it won't be enough (this article)
Mother Jones17 Nov

OpenAI announced on Monday that four independent assessors will evaluate the safety and effectiveness of its AI models. However, critics argue that the assessors lack full access to OpenAI’s technology and that the company retains control over which staff members the assessors can interact with. Previous assessments faced limitations, such as short on-site investigation times and reliance on potentially unreliable…

OpenAI’s “Independent Assessment” Doesn’t Look Independent at All →

Compared with similar articles

This article reads leans left and hedges 13% of its claims. Each row says how that neighbour differs.
New York Post
Hedges 4 of 23 claims Outlet record 66/100
“Article A reports on the lawsuit filed by California Attorney General against OpenAI, while Article B discusses OpenAI's announcement of a new AI safety program.”
Washington Examiner
Leans left Hedges 0 of 2 claims Outlet record 67/100
“Article A describes a lawsuit filed by California against OpenAI, while Article B reports on a new AI safety program announced by OpenAI. These are distinct events occurring at different times.”
Al Jazeera
Leans right further right than this Hedges 3 of 27 claims Outlet record 60/100
“Article A reports on California filing a lawsuit against OpenAI over alleged lack of proper safeguards, while Article B discusses OpenAI's announcement of a new AI safety program and concerns raised by researchers.”
NPR
Leans strongly left further left than this Hedges 0 of 19 claims Outlet record 60/100
“Article A reports on California suing OpenAI over hacking risks, while Article B discusses OpenAI announcing a new AI safety program. These are different events occurring on different dates.”
Mother Jones
Leans strongly left further left than this Hedges 2 of 21 claims Outlet record 60/100

Story

OpenAI's AI Safety Program Criticized
Technology · 2 article(s) covering the same event. See how they differ ↑
How this article was measured signals, leaning charts, publisher record, byline, topics and claims

Signals How these are calculated →

Claims extracted
31
sentences that state something as happening or true
Uncertain
13%
4 of 31 hedged ("reportedly", "may", "according to")
Leaning
Leans left
of the writing, not the subject · beta estimate
Publisher signals
60/100
NPR's corrections and hedging, across everything collected from it. Not a measure of accuracy.
Outlets on this story
2
Technology
Same framing elsewhere
1
articles that frame the event the way this one does
Analyzed 2026-11-18 · how these are computed

How this is being covered How these are calculated →

Article leaning vs. publisher reliability
Source leaning vs. consistency

Publisher

NPR · 1381 article(s) · 5 correction(s) detected
Running correction rate · 5 correction(s)

Who wrote this

Huo Jingnan
6 article(s) here · 1 carrying a prediction
The company announced Monday that staff from four AI safety and research organizations will become "embedded assessors" to "give people a clearer view of how labs are evaluating and managing [potential] risks.
In a statement to NPR, OpenAI spokesperson Drew Pusateri said the lawsuit "appears to single out OpenAI over a challenge that's been identified by every major AI lab" and that the company has "engaged—and will continue to engage—constructively with the Attorney General's office to share information and advance strong, practical safety standards for the entire industry."
2026-11-16 · assertive framing · California sues OpenAI over hacking risks
The stakes in this race are high – it's one of the competitive U.S. House races that could determine control of the chamber.
They also expressed worries the company may walk back a recent safety commitment.
And I think that is something that a few years ago, an Iranian operation like this would have struggled a lot more with."
Why are the people building the most powerful AI so worried about what it could do?
Also by Huo Jingnan
Nothing else under this byline is closely related to this article, so these are simply their most recent.
All 6 articles by Huo Jingnan →

Topics

Altman Anthropic Hugging Face OpenAI Semafor

Subjects

OpenAI ORG · 9× Altman PERSON · 3× Anthropic ORG · 2× Von Arx PERSON · 2× Apollo Research ORG · 1× Lama Ahmad PERSON · 1× Nightingale ORG · 1× Sam Altman PERSON · 1× Semafor ORG · 1× Sydney Von Arx PERSON · 1×

Narrative

One recent report about Anthropic using AI to significantly speed up model development has prompted concerns in the field about how close the top companies are to using AI to autonomously build even more powerful models, a stage known as recursive self-improvement.
framing: assertive · carried by 1 article(s) · first seen 2026-11-18
The company announced Monday that staff from four AI safety and research organizations will become "embedded assessors" to "give people a clearer view of how labs are evaluating and managing [potential] risks.

Claims (31 extracted, 4 hedged)

OpenAI announces new AI safety program; researchers worry it won't be enough AI safety researchers say OpenAI's new third party evaluator program falls far short of what the company promised, raising concerns about its effectiveness. asserted
company → announce → effectiveness
The company announced Monday that staff from four AI safety and research organizations will become "embedded assessors" to "give people a clearer view of how labs are evaluating and managing [potential] risks. asserted
labs → announce → risks
" "Getting an outside opinion on the safety work and the robustness of these systems is going to be increasingly important and valuable," CEO Sam Altman said at Semafor's Silicon Valley and the World conference on Monday. asserted
Altman → get → Monday
However, many AI safety researchers say the new program is a far cry from Altman's vow in September that independent evaluators would have "employee-like access" to internal company systems. asserted
evaluators → say → systems
Altman's promise came after details of the Hugging Face attack by OpenAI's autonomous systems shook the AI field. asserted
details → come → field
Worries of the technology's catastrophic risk captured public attention, and rival Anthropic said it would set up such a program. asserted
it → capture → program
"We are giving deep access to some of the most sensitive tools and models our researchers use...that are the most important for assessing key, relevant and timely safety claims," said Lama Ahmad, OpenAI's frontier assurances lead, on X. uncertain
Ahmad → give → X.
Prior to these new embedded evaluation programs, AI companies have brought in outside researchers to assess the capabilities of AI models and identify safety loopholes the companies have missed. asserted
companies → embed → loopholes
The work of outside evaluators have already helped shed light on the companies' work, and have raised public awareness of safety risks that the technology poses. asserted
technology → help → that
One recent report about Anthropic using AI to significantly speed up model development has prompted concerns in the field about how close the top companies are to using AI to autonomously build even more powerful models, a stage known as recursive self-improvement. asserted
companies → use → improvement
Some in the field worry that if this happens, the velocity of development could lead to catastrophic consequences for humanity and are calling for new regulation against the practice. uncertain
velocity → worry → practice
Researchers say the actual announcements of how these programs will work suggests something more limited than what Altman promised. uncertain
Altman → say → what
"The assessors have very narrow scopes to their access, and only two of them seem to really be allowed to talk to employees," said Sydney Von Arx, founder of AI safety research group Nightingale. asserted
Arx → have → Nightingale
One of the announced assessors, Apollo Research, outlined its limited access to OpenAI employees. asserted
One → announce → employees
If employees that haven't been whitelisted, or pre-approved, reach out to the embedded assessors, then the assessors will have to loop OpenAI in before having substantive conversations with the employees. asserted
assessors → whiteliste → employees
It's important for outside researchers to be able to interview employees, Von Arx said. asserted
Arx → interview → employees
"They help you find your blind spots. asserted
you → help → spots
They tell you what you're missing." asserted
you → tell → what
Another glaring omission: "Nobody has access to internal Slack," Von Arx said, referring to the corporate messaging tool, which would allow assessors to see what employees are talking about with one another. asserted
employees → have → one
In a statement to NPR, OpenAI said there's currently no standards for embedded evaluations. asserted
OpenAI → say → evaluations
"We expect to deepen these relationships over time and share what we learn to help establish best practices for effective embedded assessments across the industry," OpenAI spokesperson Gaby Raila said. asserted
Raila → expect → industry
The company also said assessors can ask to talk to employees that haven't already been whitelisted, and that there is a process in place to get statements from OpenAI about things they've learned informally. asserted
they → say → things
The third-party organizations will focus on various aspects of AI safety, according to OpenAI. uncertain
organizations → focus → OpenAI
Some of the work is related to preventing models from trying to engage in undesirable behavior, and other parts are related to catching bad behavior when it happens. asserted
it → relate → behavior
The assessors will have access to some internal models, training processes and some OpenAI employees. asserted
assessors → have → models
The assessors will be able to publish their findings after OpenAI redacts confidential information, the company said. asserted
company → publish → information
OpenAI's commitment to embedded evaluations was questioned when three employees who worked on safety were fired. asserted
who → embed → safety
They subsequently said they were worried OpenAI is walking back from its promise of having embedded evaluators. asserted
OpenAI → say → evaluators
One of the fired employees, Tomek Korbak, was OpenAI's technical point of contact for two outside evaluators that investigated the Hugging Face attack. asserted
that → fire → attack
"Embedded auditing will only fulfill its aims if the auditors really get employee-level access as opposed to being hired to do a specific job," he posted on X in response to OpenAI's announcement. asserted
he → fulfill → announcement
"Sadly, OpenAI makes it clear now that [Sam Altman's] September tweet about 'employee-like access' was just an awkward overpromise that the company wants to quietly walk back." asserted
company → make → that
Give feedback · no sign-in needed

No sign-in needed for now. Your note is sent without a name, so it cannot be answered.

History Support