Skip to content
You can now search across every topic, entity and event.What's new
AI: Jobs, Power & Money
27JUL

Anthropic withholds Mythos from public release

2 min read
10:02UTC

Anthropic's most capable model scored 83.1% on vulnerability reproduction but will not be released publicly, going instead to twelve partners through a $100 million restricted programme.

← Back to SAP freezes R&D headcount as others deny AIJump to analysis ↓
EconomicDeveloping
Key takeaway

Anthropic set the precedent for withholding a frontier model from public release over systemic risk.

Anthropic released Claude Mythos Preview exclusively to twelve partner organisations through Project Glasswing on 8 April 2026, backed by $100 million in model usage credits 1. The model autonomously identified thousands of zero-day vulnerabilities across every major operating system and browser, including a 27-year-old OpenBSD flaw that had survived five million automated tests. On the CyberGym benchmark it scored 83.1% on vulnerability reproduction, compared with 66.6% for Anthropic's previous top model.

Anthropic has explicitly stated it will not release Mythos to the public. The twelve Glasswing partners include AWS, Apple, Google, Microsoft, CrowdStrike, Palo Alto Networks, and JPMorgan. Goldman Sachs, another partner, published displacement research the same week showing AI substitutes 25,000 jobs per month , placing these institutions on both sides of the AI capability and labour displacement story.

A Tom's Hardware review challenged the marketing: the "thousands" claim rested on only 198 manual reviews, and many flagged flaws were in outdated software 2. The Bessent-Powell emergency meeting suggests federal regulators took the risk seriously regardless.

Deep Analysis

In plain English

Every major AI model so far has been made available to the public, either free or via subscription. Anthropic has broken that pattern. Claude Mythos Preview is restricted to twelve partner organisations, including tech giants and financial institutions. The model can automatically find previously unknown security flaws in software at a scale that has never been seen from an AI system. Anthropic's position is that the capability is powerful enough that releasing it widely would create unacceptable risk, the same software that makes it useful to defenders could be used by attackers. So it is being distributed under a controlled programme called Project Glasswing, with $100 million in subsidised usage credits. A technical review by Tom's Hardware found that some of the specific claims were overstated, but the underlying capability gap the model represents is real.

Deep Analysis
Root Causes

Anthropic's capability assessment rests on the CyberGym benchmark jump from 66.6% to 83.1%, a 16.5-percentage-point improvement in autonomous vulnerability reproduction. Anthropic's own research on 'observed exposure' shows computer programmers face 75% task coverage from Claude-class models; Mythos's security capabilities represent the first instance where the coverage figure has operational implications beyond individual productivity.

The restriction decision reflects Anthropic's founding premise that AI safety and capability development must remain linked. Project Glasswing's $100 million credit allocation is structured as a subsidy for defensive deployment, not a commercial launch, which is itself novel for a frontier model.

What could happen next?
  • Precedent

    The first frontier AI model explicitly withheld from public release establishes capability-gating as a legitimate deployment option for safety-constrained AI systems.

    Medium term · 0.85
  • Risk

    The twelve Glasswing partners, which include both defensive (CrowdStrike, Palo Alto) and dual-use (AWS, Google, Microsoft) organisations, may deploy Mythos capabilities in ways beyond Anthropic's stated defensive intent.

    Short term · 0.62
  • Opportunity

    Security professionals who can interpret and direct AI-identified vulnerability data at scale represent a new premium-tier role the model creates even as it automates routine scanning.

    Medium term · 0.71
Sources:Anthropic
Mentions:Goldman Sachs →Carnegie Endowment for International Peace →Prima →Project Glasswing →Anthropic →Claude Mythos Preview →Amazon →Apple →Google →Microsoft →CrowdStrike →Palo Alto Networks →CyberGym →RUSI →JPMorgan Chase →
First Reported In

Update #5 · The model they won't release

Anthropic· 10 Apr 2026
Read original →
Causes and effects
This Event
Anthropic withholds Mythos from public release
The first frontier AI model deliberately withheld from public deployment establishes a precedent for capability-gated release in the industry.
Led to
Anthropic drops ASL, expands Glasswing partners
Anthropic's 7 April system card and Glasswing expansion directly extended the Glasswing programme launched on 8 April with the original twelve partners.
Occurred 7 Apr 2026
Read story →
Dimon: JPMorgan displaced workers from AI
JPMorgan is one of the original twelve Glasswing partners, creating the contradiction of a bank attending emergency Treasury meetings about AI risk while internally displacing staff through AI deployment.
Occurred 24 Feb 2026
Read story →
Different Perspectives
European Commission
European Commission
The European Commission's draft Annex III guidelines, closed for comment on 23 July, treat algorithmic scoring in recruitment, pay and termination as high-risk regardless of whether a human signs off, echoing Spain's Audiencia Nacional ruling 101/2026 on concealed scheduling algorithms. Brussels is shifting the fight from counting AI job losses to assigning legal liability for the tools themselves.
Office for National Statistics
Office for National Statistics
The Office for National Statistics recorded UK vacancies rising to 712,000 on 21 July, the first quarterly increase this beat has tracked, with payrolled employment down 85,000 on the year against May's 210,000 fall. The bulletin names no AI cause anywhere, and that is the point: nothing in the release confirms the displacement story it gets cited to support.
Christian Klein, SAP
Christian Klein, SAP
Christian Klein told investors on 23 July that SAP's research headcount will not grow for twelve months because AI agents and their token costs are absorbing the work, not because SAP is cutting jobs. He frames it as commercial arithmetic: the cost of AI-assisted coding tokens plus the salaries specialist AI hires command, not people being replaced by machines.
Betsey Stevenson, University of Michigan
Betsey Stevenson, University of Michigan
Betsey Stevenson argued that the 187,000 jobless-claims reading describes a market that hires little and fires little, not one AI is emptying. She said the real damage hides in eligibility rules and suppressed job postings, not in the headline layoff counts employers keep denying.
Comisiones Obreras, UGT and Concentrix's A Coruña works committee
Comisiones Obreras, UGT and Concentrix's A Coruña works committee
Comisiones Obreras, UGT and Concentrix's A Coruña works committee blamed Microsoft's push toward AI self-service for the 80 redundancies unions signed off on 22 July, not unavoidable business cause. A second Coruña procedure covering 80 more jobs runs to a 31 August deadline, and the unions want the state, not the employer, setting the pace of AI-driven cuts.
Stanford's 'We Must Act Now' signatories
Stanford's 'We Must Act Now' signatories
More than 200 academics, including 16 Nobel laureates, published a 13 July letter warning of AI-driven labour disruption, citing Daron Acemoglu's NBER estimate that AI's total factor productivity gain stays under 0.66% over ten years. The letter's own cited economics sit well below Goldman Sachs Research's 1.5-percentage-point estimate published the same week.