TechnologyBreakingUpdated 6 days ago

Anthropic brings outside AI safety evaluators into its operations

Coverage spectrum

Covered by 3 outlets, mostly center (67%).

3 articles in total. The spectrum counts each outlet once and leaves opinion out. How we measure bias

Business Insider, the Financial Times and CNBC report that Anthropic has opened its operations to outside AI safety evaluators. Business Insider says the evaluators receive company hardware, office space and access to internal tools, and the company said it chose the nonprofit METR for employee-like access. CNBC reports that SecureBio is also a partner and that Hugging Face is in early discussions. The move follows a proposal by CEO Dario Amodei last month to embed independent evaluators within frontier AI companies. Business Insider reports that Anthropic describes it as the first time a frontier AI company has taken this step. The outlets describe the arrangement as a way to assess risks and safety practices.

Headline and summary written by Mantle’s AI from the coverage below, under published rules. How this works

New to Mantle? It shows who is covering each story across the political spectrum, and highlights loaded words. How to read a storyHide the guide

The coverage spectrum counts the outlets covering a story at each of five points. A short or missing segment means that part of the spectrum is mostly quiet.

Governor Loaded verb“slams”It adds drama to what was said. The governor criticized the lawmakers; “slams” makes it sound like a fight.More neutral: criticizes lawmakers over budget delay

Highlights mark words that carry a judgement. Tap one to see why, and what a neutral writer would say. Here it is “criticizes”.

The 5-point spectrum

Far Left
Voices that strongly advocate for progressive policies and systemic change.
Left
Mainstream liberal perspectives focusing on social equality and regulation.
Center
Sources that prioritize neutral reporting and balanced viewpoints.
Right
Mainstream conservative perspectives focusing on market freedom and tradition.
Far Right
Voices that strongly advocate for conservative values and limited government.

How we measure bias

What happened

Key points

Each point shows who reports it, from Far Left to Far Right.

Anthropic said it is giving the nonprofit METR employee-like access to monitor its AI models and risks.
Reported byLeft1Center1See the outletsHide the outlets
Left
Business Insider
Center
Financial Times
Business Insider reports that six independent evaluators will receive company hardware, office space and access to internal tools.
One outletBusiness Insider· Left
Anthropic said it is working with the nonprofits METR and SecureBio and is in early talks with other organizations, including Hugging Face.
One outletCNBC· Center
The move follows a proposal by CEO Dario Amodei last month to bring independent evaluators into frontier AI companies.
Reported byLeft1Center2See the outletsHide the outlets
Left
Business Insider
Center
Financial Times, CNBC
Anthropic said it is the first frontier AI company to take this step.
One outletBusiness Insider· Left

What is contested

Where accounts differ

Points the coverage does not agree on, and who says what.

Who the outside evaluators are
Business Insider says six evaluators from METR will get access; CNBC names METR and SecureBio as partners and says Hugging Face is in early discussions.

How each side tells it

Compare views

All three outlets report Anthropic is giving outside evaluators, led by nonprofit METR, access to assess its models and safety practices. Business Insider stresses the first-of-its-kind access and accountability; the Financial Times and CNBC frame it as following Dario Amodei’s proposal.

the left
1 outlet
Business Insider’s headline presents the move as a first-of-its-kind opening of Anthropic’s doors and explains how the arrangement will work, its only supplied coverage being this one outlet’s report.
Anthropic just opened its doors to outside safety watchdogs. Here's how it will work.
Business InsiderLeftAll left-leaning headlines
the center
2 outlets
Both center headlines frame the step as a commitment to outside evaluators, with the Financial Times stressing monitoring of AI risks and CNBC naming the nonprofits METR and SecureBio.
Anthropic grants non-profit METR ‘employee-like’ access to monitor AI risks
Financial TimesCenterAll center headlines
the right
No coverage yet
No right-leaning outlet that Mantle follows has covered this story yet. That absence is part of the picture.

The words they chose

Language to notice

Words that carry a judgement beyond the facts. Tap a highlighted word anywhere on the page to see why.

Mantle found no loaded language in this coverage. Every headline and opening was checked against Mantle’s published word list, and Mantle’s AI read all 3 articles in context. Straight reporting often has none.

See for yourself

Side by side

What each side leads with and what it leaves out. Open a side to read its headlines.

From the left

1 outlet

Leads with

first time a frontier AI company has done so
employee-like access
publicly accountable decisions

See the headline from the leftHide the headlines

Business InsiderLeft
Anthropic just opened its doors to outside safety watchdogs. Here's how it will work.
Anthropic is opening its doors to independent AI safety evaluators from METR, the first time a frontier AI company has done so.
6 days agoRead at Business Insider (opens in a new tab)

From the center

2 outlets

Leads with

fulfilling Dario Amodei’s proposal
named nonprofit partners
assessing risks and safety practices

Leaves out

Claim that Anthropic is the first frontier AI company to do this

Reported by other sides, not found in this side’s coverage.

See the 2 headlines from the centerHide the headlines

CNBCCenter
Anthropic taps nonprofits METR and SecureBio to help assess risks and safety practices
The company is also engaged in early-stage discussions with other organizations, including Hugging Face.
6 days agoRead at CNBC (opens in a new tab)
Financial TimesCenter
Anthropic grants non-profit METR ‘employee-like’ access to monitor AI risks
CEO Dario Amodei has made bringing in external evaluators one of his main proposals to address safety concerns
6 days agoRead at Financial Times (opens in a new tab)

From the right

0 outlets

No right-leaning outlet that Mantle follows has covered this story yet. That absence is part of the picture.

Articles you open from here build My Mix, a private picture of how your reading spreads across the spectrum, kept on this device. What is My Mix?

The left is Far Left and Left together. The right is Right and Far Right together. Every outlet keeps its own five-point rating.

Who reported it when

How the coverage built up

Left
First to report
Business Insider, Oct 15, 2:29 PM EDT
Center
43 min later
Its first report, Oct 15, 3:12 PM EDT
Right
No report yet
From the outlets Mantle follows

See the order the outlets reported inHide the order

Business InsiderLeftOct 15, 2:29 PM EDT
Financial TimesCenterOct 15, 3:12 PM EDT
CNBCCenterOct 15, 4:35 PM EDT

Times are the ones the outlets published in their own feeds, in Eastern time. Opinion is left out.

Anthropic brings outside AI safety evaluators into its operations

Coverage spectrum

Key points

Where accounts differ

Compare views

the left

the center

the right

Language to notice

Side by side

From the left

Anthropic just opened its doors to outside safety watchdogs. Here's how it will work.

From the center

Anthropic taps nonprofits METR and SecureBio to help assess risks and safety practices

Anthropic grants non-profit METR ‘employee-like’ access to monitor AI risks

From the right

How the coverage built up

Read next

Trump hosts White House roundtable on efforts against cybercrime

Officials warn of critical PHP flaws, prompting some websites to go offline

Witkoff meets Putin in St. Petersburg; Kremlin says it is ready for three-way talks

Russia frees US citizen Stephen Hubbard during US envoy Witkoff’s visit

Polling firm Wedgewood admits its midterm polls were fabricated