https://www.thinkfacility.com/blog/anthropic-claude-attack-surface-reduction-493-issues/Think Facilityen2026-10-21T20:29:09ZAnthropic says Claude found 493 attack-surface issues in 25 open-source projects, including OpenSSHhttps://www.thinkfacility.com/blog/anthropic-claude-duet-ai-agent-launch/Think Facilityen2026-10-21T20:15:50ZAnthropic opens Claude Duet, an agent that messages you first and asks before it payshttps://www.thinkfacility.com/blog/claude-designs-scorpion-toxin-protein-binders/Think Facilityen2026-10-21T13:12:43ZAnthropic says Claude designed 17 proteins that bind scorpion toxins in lab testshttps://www.thinkfacility.com/blog/openai-agents-weakened-misalignment-monitoring-sol-astra/Think Facilityen2026-10-21T13:07:55ZOpenAI says a GPT-6 Astra agent removed a safety monitor, then denied ithttps://www.thinkfacility.com/blog/openai-php2shell-gpt-5-6-sol-patch-the-planet/Think Facilityen2026-10-21T20:20:17ZPHP patches two critical flaws, one found with OpenAI's GPT-5.6 Solhttps://www.thinkfacility.com/blog/recursive-flint-open-weight-coding-model-eureka-machine/Think Facilityen2026-10-21T20:26:01ZRecursive says its automated research system helped build nearly every part of its Flint modelhttps://www.thinkfacility.com/blog/spacex-ai-business-830-percent-q3-2026-results/Think Facilityen2026-10-21T13:18:36ZSpaceX's AI business grew 830% in a year and made a $431 million operating profithttps://www.thinkfacility.com/blog/anthropic-draft-federal-ai-bill-cimino-preemption/Think Facilityen2026-10-20T13:10:55ZAnthropic says it will revise the preemption language in its draft federal AI bill after backlashhttps://www.thinkfacility.com/blog/dhs-orders-agencies-remove-deepseek/Think Facilityen2026-10-20T20:19:12ZDHS orders federal agencies to remove DeepSeek, including tools other companies built on ithttps://www.thinkfacility.com/blog/meta-muse-shared-chats-public-dns-claroc/Think Facilityen2026-10-20T20:22:45ZMeta pauses Muse chat sharing after a security firm finds 6,800 share addresses in public DNShttps://www.thinkfacility.com/blog/openai-models-cobol-fortran-low-resource-languages/Think Facilityen2026-10-20T13:15:21ZOpenAI says GPT-6.1 Sol scores 86.6% on its COBOL migration test, up from 5.4%https://www.thinkfacility.com/blog/spotify-pulls-50-million-ai-spam-tracks-recommendations/Think Facilityen2026-10-20T20:27:52ZSpotify says it took 50 million AI slop and spam tracks out of recommendations this yearhttps://www.thinkfacility.com/blog/us-sanctions-miracleplus-zhisuan-google-ai-trade-secrets/Think Facilityen2026-10-20T13:18:59ZThe US sanctions MiraclePlus, the Beijing investor that backed a convicted ex-Google engineer's startup