Anthropic launches a cybersecurity initiative
It is working with 11 infrastructure defense partners to use AI against cyberattacks.
Tip
Why AI companies are moving into security, and whether ordinary users will notice
StepFun releases Step 5 preview
A 600B-parameter mixture-of-experts model with a 1-million-token context window.
Tip
How much material “1 million tokens” actually holds
Anthropic updates its usage policy
It bans sustained cruelty toward Claude and deceptive election campaigns.
Tip
What AI usage policies say, and which rules users should know
Liquid AI open-sources two decision models
d1-3B and d1-omni-600M, built on a “zero output token” architecture.
Tip
Where small models that only make decisions, not long text, are useful
AI alignment platform Arena raises funding
A 3.1B valuation.
Tip
What “alignment” means and why investors are paying for it
⚠️ Compiled by an AI assistant from public AI news roundups; original articles are not individually verified. For reference only.