Anthropic
Claude models and Claude Code, enterprise partnerships and research. Every brief item about Anthropic, newest first, each with its primary source.
Where it stands
Almost everything Anthropic did this week is about handing capabilities over while writing the boundaries down more clearly. On October 6 it expanded the Cyber Verification Program into three tiers; on October 7 it released Claude Haiku 5.5 ($0.10 per million input tokens under 100K, about 75% cheaper than Haiku 4.5 on average); and on October 8 it shipped two things at once: the Cyber Mission (11 founding partners for critical infrastructure defense, free model scans for open-source projects) and a new Usage Policy effective November 12.
Capability distribution and rules are moving together. On defense, the model produces vulnerability reports with a proof of concept and suggested fix, sent without human review, with an expected true-positive rate above 90%. On rules, the new policy is more specific about Claude controlling physical hardware, recommendations in high-risk domains, and unsupported regions defined by ownership. A cheaper Haiku 5.5 drives call volume up; the rules decide which calls are allowed.
What to watch next: OSS Scanner's false positives on real projects, how fast CIDP grows beyond its first cohort, and how the new policy is enforced on hardware, hiring and finance apps after November 12.
My advice: before November 12, check whether your product falls in a high-risk domain and add a human review path and an AI-generated notice where needed; on cost, keep tiering by task, moving high-volume short calls to Haiku 5.5 and leaving complex coding to Sonnet 5.5 or Opus 5.5.
Darius · Updated Oct 9, 2026
Timeline 5 items
-
Oct 9 · Fri · 2 items
- Safety anthropic.com ↗Anthropic launches its Cyber Mission: 11 founding partners for critical infrastructure defense and free model-run vulnerability scans for open-source projects
Take · If you maintain a widely used open-source library, I'd sign up for OSS Scanner, but assign one person to triage the reports first: no human review and roughly 90% true positives means about one in ten may be noise. I'll start by looking at report quality on the upload and parsing libraries AI Cloud Drive depends on.
From the Oct 9 brief · item 01 → - Safety anthropic.com ↗Anthropic updates its Usage Policy, effective November 12: new rules for controlling physical hardware and region limits that follow ownership
Take · AI Interview falls squarely in the employment category, so before November 12 I'll make sure of two things: the results page says which assessments come from AI, and the HR side keeps a human review and override path. Teams with offshore entities or foreign investors should also check whether their ownership structure runs into the regions clause.
From the Oct 9 brief · item 02 →
-
-
Oct 8 · Thu · 1 item
- Model release anthropic.com ↗Anthropic releases Claude Haiku 5.5 at $0.10 per million input tokens under 100K, about 75% cheaper than Haiku 4.5 on average; Sonnet 5.5 cache reads cut in half
Take · I'm moving short-prompt calls, such as PandaClaws' classification, summaries and rewrites and AI Cloud Drive's document Q&A, onto Haiku 5.5 for a comparison run right away: under 100K tokens it costs a tenth of the old price, so a big model planning and Haiku doing the subtasks finally pencils out. The new tokenizer uses slightly more tokens, so compare cost per completed task rather than unit price; the cost table in my post on what it costs to keep three AI products online needs redoing too.
From the Oct 8 brief · item 01 →
-
-
Oct 7 · Wed · 1 item
- Safety anthropic.com ↗Anthropic expands its Cyber Verification Program into three tiers of access to advanced cyber capabilities on Opus 5.5, Sonnet 5.5 and Mythos 5.1
Take · If your team keeps getting blocked when using Claude for code audits or security triage, this is the proper route: apply for the Defense tier first, since it's reviewed fastest. Note that enrolling means data retention, so for products handling private user files like my AI Cloud Drive, run security testing against a scrubbed copy rather than real user data.
From the Oct 7 brief · item 03 →
-
-
Oct 5 · Mon · 1 item
- Industry anthropic.com ↗Anthropic commits $100M to Claude Frontier Academy, aiming to train 10,000 Frontier Deployed Engineers by end of 2027
Take · When a model vendor starts training the people who wire models into businesses, the enterprise bottleneck clearly isn't capability but integration and delivery. For independent developers and small teams, a track record of shipping real use cases is worth more than a badge, and that's where the work is.
From the Oct 5 brief · item 04 →
-