HaiAI123

Curated Global AI Tools Directory

RisingIndustry update
↑ 25%

OpenAI launches first phase of embedded evaluations with four external research groups

11/17/2026 — 11/17, 08:19·6 sources (all)·6 reports (all)·In the 2026-11-17 briefing

Key takeaways

  • OpenAI has launched the first phase of its embedded evaluation program, partnering with Apollo Research, METR, Transluce, and Goodfire to in…

Background & analysis

On November 16, 2026, OpenAI announced via OpenAI News that it was launching the first phase of its embedded assessments, working with four external organizations, Apollo Research, METR, Transluce, and Goodfire, to independently evaluate model alignment, safeguards, and monitoring systems. The announcement did not specify which models would be assessed, the scope of the work, or any timeline. On November 17, OpenAI confirmed that it had signed agreements with the four independent assessors for this first phase, though it still did not disclose which models were covered or when the work would unfold.

Reports published on November 17 added detail on two of the partners. According to TechCrunch, Apollo Research will embed six safety evaluators inside OpenAI as a pilot to investigate whether its models try to undermine the alignment and control mechanisms of their own or successor models, a behavior researchers call scheming, and to check whether safeguards can detect it; this evaluation is referenced in the o1 system card. The pilot start date has not been made public, and the report noted that restrictions on access and disclosure could affect the evaluators' independence. A separate report from The Next Web stated that the six Apollo Research evaluators will work inside OpenAI for four months, examining whether models attempt to weaken alignment and control measures during training, testing, and internal use. OpenAI will control staff access and may redact some evidence, but it cannot conceal Apollo's verdicts or withhold material simply because findings are unfavourable.

Regarding Goodfire, IT之家, citing The Information, reported that OpenAI had previously explored acquiring Goodfire for more than $5 billion but did not complete the deal. Goodfire said it will station interpretability researchers inside OpenAI, using its own interpretability analysis platform to monitor model training, flag unintended behavior, and support model forensics. These researchers will receive deep access to OpenAI's internal models and systems, and results will primarily go to OpenAI's internal research teams. The specific roles of METR and Transluce in this phase have not been disclosed in the available reports.

AI-generated from 6 reports · updated 3 hours ago

Latest turnOpenAI has announced partnerships with Apollo Research, METR, Transluce, and Goodfire to launch the first phase of its Embedded Assessments program, focused on model interpretability and safety evaluations. The Information reports that OpenAI had explored acquiring Goodfire for more than $5 billion but no deal was reached. Goodfire researchers will embed at OpenAI to monitor training and detect unexpected behavior.

updream
24-hour heatHeat index 445 · peak 494 · 3h ago
24 hours agonow

Reports on this story headlines open the original

Today
  1. OpenAI has announced partnerships with Apollo Research, METR, Transluce, and Goodfire to launch the first phase of its Embedded Assessments program, focused on model interpretability and safety evaluations. The Information reports that OpenAI had explored acquiring Goodfire for more than $5 billion but no deal was reached. Goodfire researchers will embed at OpenAI to monitor training and detect unexpected behavior.

    IT之家AI score 56
  2. OpenAI is working with Apollo Research to evaluate whether its models could covertly pursue goals that conflict with developer intent, a behavior researchers call scheming, and to test whether safeguards can detect it. For teams building AI workflows, the practical takeaway is that model outputs alone are not enough when a system can access data, tools or operational processes.

    DEV Community · AIAI score 62
  3. According to The Next Web, Apollo Research will place six evaluators inside OpenAI for four months to check whether its models try to undermine alignment and control measures during training, testing and internal use. OpenAI controls staff access and can redact some material, but it cannot hide Apollo's verdicts or withhold evidence just because they are unfavourable.

    The Next WebAI score 60
  4. Apollo Research said it will embed six safety evaluators inside OpenAI as a pilot to investigate whether its models try to subvert their own or successors' alignment and control mechanisms. The start date is undisclosed, and TechCrunch notes that limits on access and disclosure raise questions about the evaluators' independence.

    TechCrunch · AIAI score 71
  5. OpenAI said it has signed agreements with four independent assessors — Apollo Research, METR, Transluce and Goodfire — for the first phase of its embedded assessments. The announcement did not specify which models will be assessed, what the scope covers, or a timeline.

    TechmemeAI score 53
Yesterday
  1. OpenAI said it has started the first phase of embedded assessments, working with Apollo Research, METR, Transluce, and Goodfire to independently evaluate model alignment, safeguards, and monitoring systems. The announcement did not detail the scope of the evaluations or their results.

    OpenAI NewsFirst-partyAI score 51

Other stories people are talking about

How is heat calculated?About the method

Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.

This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.

Surge
Discussion rising fast
New
First report within 6 hours
Rising
Still gathering discussion

Back to AI industry updates →