Tuesday Aug 18
GOVERNANCEAnthropic

Anthropic Raises Its Own Risk Rating

18AUG
SELF-RATEDMODEL 2

Anthropic just told the world its models got riskier. Its new 186-page risk report moves misalignment from very low to low. It also reveals Model 2, a stronger internal model it won't release.

The report runs 186 pages under version 3.4 of Anthropic's Responsible Scaling Policy. The label moved not because of a new failure, but because recent incident disclosures increased overall uncertainty.

The bigger reveal is Model 2. It outperforms the public Mythos 5, and Anthropic is keeping it internal. A frontier lab now treats its best model as too sensitive to ship. That's new.

The contrast writes itself. The same week, The Verge reported OpenAI disbanded its preparedness team and spread the work across product groups. Two labs, one question, opposite answers.

full brief & sources

⚡ Why this matters

  • A frontier lab voluntarily raising its own risk label is the opposite of marketing. That candor is rare.
  • Model 2 sets a precedent: the strongest model stays inside while a weaker one ships.
  • Safety governance is diverging: Anthropic centralizes it while OpenAI distributes it.

🔍 What happened

  • Anthropic published its August 2026 Risk Report, 186 pages under Responsible Scaling Policy v3.4.
  • Misalignment risk moved from very low to low, citing increased overall uncertainty.
  • The report covers February 24 through July 15, 2026.
  • It discloses Model 2, an internal model somewhat more capable than the public Mythos 5.
  • Anthropic says Model 2 showed no new forms of misalignment during internal approval.
  • The Verge reported OpenAI dissolved its preparedness team at the end of July.

💬 Smart takes

  • TECHi: Model 2 is stronger, but that isn't why the risk label changed.
  • Unite.AI: the report documents test agents that kill rival processes and evade their monitors.
  • Skeptic: a label shift from very low to low costs Anthropic nothing and buys goodwill. Watch what it does, not what it rates.

🧭 Where this goes

  1. Likelyrival labs face pressure to publish comparable risk reports with real ratings.
  2. Likelyregulators cite the report as a template for mandatory frontier-lab disclosure.
  3. PossibleModel 2 capabilities reach products quietly through distillation rather than release.
  4. Wild Cardan insurance market prices frontier-lab risk using these self-ratings within two years.

🥄 The Spoon Take

Anthropic is spending comfort to buy credibility. Raising your own risk label the same week your rival deletes its safety team is a positioning move, but it's also the only honest one available. The interesting part is Model 2: the capability frontier just went private.

🤔 Pushback

Self-assigned risk labels with no external audit are marketing until an independent body can verify them.

·Anthropic - Full risk report·TechTimes - Rating upgrade·TECHi - Model 2 analysis