AI security, model, and MCP updates #134
Today's Letter
- OpenAI extends Daybreak cyber access to Ukraine
- Anthropic, Claude Opus 5.5 released
- Databricks, Genie One MCP Generally Available
OpenAI extends Daybreak cyber access to Ukraine
- OpenAI will provide Ukraine’s government access to the Daybreak program for civilian infrastructure cyber defense
- The Ministry of Digital Transformation will work with OpenAI to identify software vulnerabilities and develop and test fixes
- Daybreak supports authorized security work including legacy software review, suspicious-activity investigation, vulnerability validation, and fix testing
- CERT-UA handled nearly 6,000 cyber incidents in 2025 across hospitals, energy, telecommunications, and other essential services
- OpenAI has also provided cyber models to defenders in France, Germany, Poland, and other European countries
- CERT Polska used OpenAI models to identify six vulnerabilities in third-party router software; the vendor has released fixes
Source: openai.com
Anthropic, Claude Opus 5.5 released
- Anthropic introduced Claude Opus 5.5 on September 22, 2026, as the first model in the Claude 5.5 family
- The model matches Claude Fable 5.1 on most work while costing 40% less to run than Claude Opus 5
- API pricing is $4 per million input tokens and $20 per million output tokens; cache reads cost $0.20
- Anthropic reports gains in agentic coding, computer use, knowledge work, and long-running codebase migrations
- Opus 5.5 achieved the company’s strongest automated behavioral-audit results and improved resistance to prompt injection
- Fast mode is available in Claude Code and Claude Platform at up to 2.5x speed, with higher token pricing
- Claude Sonnet 5.5 and Claude Haiku 5.5 are scheduled to follow in the coming weeks
Source: anthropic.com
More: artificialanalysis.ai
Databricks, Genie One MCP Generally Available
- Databricks announced the general availability of Genie One MCP on September 22, 2026
- Genie One MCP brings governed business context to MCP-compatible AI assistants
- Genie Ontology defines approved metrics, data relationships, source authority, and permissions
- Integrations include ChatGPT, Claude, Microsoft Copilot, and coding agents
- Permission-aware access helps agents answer using data approved for the requesting user
- Centralized context reduces repeated schema inspection, token usage, and response latency
- Direct data access alone does not provide consistent definitions, lineage, or authoritative source selection
Source: databricks.com
More: gatesnotes.com
Jocoletter curates AI, software, and product trends for developers and builders.
#Anthropic #Databricks #OpenAI