Today: OpenAI broadens its Daybreak security program with an updated cyber model and an open-source bug-fixing push, SpaceX agrees to supply compute to open-source lab Reflection AI, and Oracle ties tens of thousands of job cuts to AI. Here are five developments worth knowing from the past few days.

OpenAI expands its Daybreak security program with GPT-5.5-Cyber and a "Patch the Planet" effort. OpenAI said on June 22 that it updated GPT-5.5-Cyber, a limited-preview model for vetted defenders, and added Codex Security tooling alongside Patch the Planet, an initiative run with Trail of Bits to help open-source maintainers find and fix vulnerabilities. The company reported the model scored 85.6% on the CyberGym benchmark, up from 81.8% for GPT-5.5, and 69.8% on SEC-bench Pro, up from 63.1%. It said the patching effort produced hundreds of identified bugs, 64 pull requests, and 51 issues across 19 projects — including cURL, Python, Go, and Valkey — in its first week, with more than 30 projects signed up. (TechCrunch) (The Hacker News) (Trail of Bits)

SpaceX signs a compute deal worth up to about $6.3 billion with open-source lab Reflection AI. Under the agreement, SpaceX will give Reflection access to NVIDIA GB300 chips at its Colossus 2 data center near Memphis, with Reflection paying $150 million per month from July 1, 2026 through 2029 and an option to terminate after the first three months. Reflection, founded by former Google DeepMind researchers and recently valued at roughly $25 billion, said the capacity will support its open-weight models. SpaceX has previously struck Colossus compute arrangements with Anthropic and Google. (CNBC) (TechCrunch) (Axios)

Oracle discloses about 21,000 job cuts over the past year and points to AI. In its annual regulatory filing, Oracle said it had reduced its workforce by roughly 21,000 employees — about 13% — over the prior 12 months, and stated that adopting AI across its operations has resulted in, and may continue to result in, workforce reductions. The disclosure was broader than previously reported figures and added to a running tally of 2026 tech layoffs in which employers cited AI. (Bloomberg) (Gizmodo) (TechCrunch)

Data-center permitting becomes a flashpoint as the AI buildout meets local opposition. CNBC reported on June 22 that the speed of permitting has become a leading factor in where companies site AI data centers, even as communities push back over electricity prices, water use, noise, and pollution. The report highlights Pennsylvania, where officials cite roughly a 72% reduction in permit processing times over the past year amid several contested projects. (CNBC)

The absence of a clear US AI framework keeps Anthropic in focus (follow-up). Building on the recent federal export action against Anthropic's top models, a CNN analysis published June 21 examined how the lack of a transparent national AI framework has left the company navigating between the government and its own safety commitments. New since our earlier coverage: President Trump told Axios in remarks published June 19 that he no longer regards Anthropic as a national security threat, and dozens of researchers and executives signed an open letter calling for a transparent, science-based risk-assessment process. (CNN)

That's today's edition of AI Current. Reply with what you're seeing in AI, and forward this to a colleague who would find it useful. See you tomorrow.