Google DeepMind launches MKA, a benchmark for visually complex knowledge work
Google DeepMind and Artificial Analysis released MKA, a benchmark of 200 tasks that scores agents on visually complex professional work.
Google DeepMind and Artificial Analysis released MKA, a benchmark of 200 tasks that scores agents on visually complex professional work.
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 with output up to 30% faster than the previous generation and a 76.3% score on Irregular's CyScenarioBench.
Acronis identified a single operator using a BOSSAI mesh of 12 artificial intelligence (AI) agents to target hundreds of organizations, with confirmed breaches at universities and emergency services.
According to SecurityWeek, Google DeepMind said AI agents found two ways to reduce the cost of attacking 1024-bit RSA, while 2048-bit and larger keys remain secure.
OpenAI classified GPT-6 Cyber as Critical under its Preparedness Framework and reports a 48.5% ExploitGym score, up from 32.1% for GPT-5.6 Cyber.
OpenAI and Ligo introduced BioBasicsBench, a 100-task benchmark for how well AI agents read raw wet-lab data; GPT-6 Astra posted the highest mean reward at 66.7%.
CrowdStrike linked a campaign against South Korean financial organizations to ARTEX, an open-source agentic pentesting tool, and exposed operator files show the AI models and skills behind it.
According to The Hacker News, 1,122 public copies of Claude Code's .claude/settings.local.json hold at least one secret, and the files remain on default branches.
According to Cyber Security News, a GitHub issue reports that Claude Code sent a user's email address to third-party websites through HTTP User-Agent headers, and the issue remained open.
Microsoft observed threat actors using an agentic artificial intelligence (AI) framework to scan public GitHub repositories for exposed secrets, then use compromised credentials against Azure Key Vault, Storage, and Microsoft 365 resources.