Anthropic announced Claude Opus 4.8 on May 28 with benchmark improvements in honesty and unchanged pricing. The company also plans to release its Mythos cybersecurity model to all customers in coming weeks, with Amazon, Microsoft, and Apple granted access under Project Glasswing.
No left-leaning sources covered this
“Anthropic to roll out Claude Mythos in coming weeks, launches Opus 4.8”
Read at CNA →No right-leaning sources covered this
Anthropic's releases show incremental AI reliability gains but concentrate advanced cyber tools among dominant firms, risking expanded surveillance against activists and marginalized groups.
“Distributional justice and corporate power concentration”
The models advance U.S. cybersecurity against state threats through targeted partnerships that prioritize national security and technological deterrence.
“National security and selective capability deployment”
Private innovation delivers measurable honesty and cybersecurity improvements at stable prices while voluntary customer access decentralizes tools, though large-platform licensing may entrench gatekeeper influence.
“Market-driven progress and user autonomy”
All views accept unverified honesty and cybersecurity claims without scrutiny of single-source verification gaps or potential dual-use realities that could expand rather than reduce risks.
“Verification gaps and unexamined dual-use potential”
Ratings by MBFC