// daily-digest · tue 23 jun 2026 · morning refresh · last 48h focus
Ten items covering automated security remediation from OpenAI, capital flows into open-source compute, enterprise AI at Samsung, AI-funded congressional primaries, and the first day of usage-credit billing for Claude Fable 5.
— refreshed for you, in 10 items.
Political committees linked to OpenAI and Anthropic collectively spent over $15 million on pro- and anti-candidate ads in competitive congressional primaries.
OpenAI releases GPT-5.5-Cyber and launches Patch the Planet with Trail of Bits and HackerOne to find and fix open-source software vulnerabilities at scale.
Reflection AI commits $150 million per month to SpaceX’s Colossus 2 data center for Nvidia GB300 compute through 2029, totaling up to $6.3 billion.
Samsung deploys ChatGPT Enterprise and Codex to more than 120,000 employees worldwide in one of OpenAI’s largest enterprise rollouts to date.
A cost breakdown of the three leading frontier AI models as Claude Fable 5 moves from subscription-included to pay-per-token billing at $10/$50 per million tokens.
Tencent begins testing Xiaowei, an AI agent embedded inside WeChat’s 1.4 billion-user interface, targeting a Q3 public rollout.
Yegge maps AI adoption culture for teams operating under the assumption that Mythos-class models are the most capable intelligence most engineers will ever access.
Willison frames prompt injection as a role-confusion failure where AI systems cannot reliably distinguish trusted principal instructions from attacker-controlled data.
Cloudflare’s new primitive lets AI agents provision short-lived accounts to access bot-protected web resources without sharing permanent credentials.
Heliconius butterflies live up to 25 times longer than close relatives and show no measurable physical decline as they age, a Nature Communications study finds.