Thursday, August 20
Boursel
Latest
Personal Finance When driving across town for cheaper gas actually pays offTech OpenAI counters Anthropic with a no-retention approach to AI safety monitoringEconomy Trump pauses 50% tariffs as the US and Canada close in on a trade dealCompanies SpaceX sought to buy AI coding startup Cognition, Bloomberg reports; its CEO says noTech OpenAI slows some frontier AI work after a model nears a critical cyber thresholdCrypto Bitcoin tops $69,000, wiping out roughly $1.7 billion in short betsMarkets Gold jumps to its highest since early June as the Treasury doubles bond buybacksPersonal Finance Treasury moves to deny refundable tax credits to immigrants without qualified statusPersonal Finance When driving across town for cheaper gas actually pays offTech OpenAI counters Anthropic with a no-retention approach to AI safety monitoringEconomy Trump pauses 50% tariffs as the US and Canada close in on a trade dealCompanies SpaceX sought to buy AI coding startup Cognition, Bloomberg reports; its CEO says noTech OpenAI slows some frontier AI work after a model nears a critical cyber thresholdCrypto Bitcoin tops $69,000, wiping out roughly $1.7 billion in short betsMarkets Gold jumps to its highest since early June as the Treasury doubles bond buybacksPersonal Finance Treasury moves to deny refundable tax credits to immigrants without qualified status

Topic: anthropic

The Pioneer Building in San Francisco, OpenAI's longtime headquarters
Tech
Wednesday, 8/19, 10:36 PM

OpenAI counters Anthropic with a no-retention approach to AI safety monitoring

  • OpenAI is previewing Private Safety Processing, a system that scans for patterns of misuse across customer sessions using automated agents instead of human reviewers, without retaining the underlying data.
  • The design is a direct answer to Anthropic's 30-day retention policy for its most capable models, and a sign that privacy has become a competitive weapon in enterprise AI.
Servers in a data center
Tech 52 d

A Chinese AI Lab Claims It Matches Anthropic's Restricted Cyber Model

  • Z.ai, one of China's leading AI startups, says its new open-source model rivals Anthropic's Mythos at finding software vulnerabilities — the very capability that got Mythos walled off from foreign users on US national-security grounds.
  • The claim rests on a thin, heavily caveated benchmark, but the point lands: export controls do little if an open alternative exists.

In brief

anthropic — news and stories on this topic

This page collects all of Boursel's coverage of anthropic. Follow the topic from one place: new stories appear here automatically as more is published. That way you stay on top of how the situation develops.

Looking for the bigger picture? Browse markets, the economy, companies and crypto, or head back to the front page for the day's top stories.