Daily AI news, sourcedOfficial news, first rumor reports and substantive progress. Each entry keeps the information published at that time.
- OpenAI makes Daybreak Red available in AWS GovCloud (US-East) via Amazon Bedrock
OpenAI says Daybreak Red is now available to approved customers in AWS GovCloud (US-East) through Amazon Bedrock, bringing GPT-5.6 Cyber for authorized cybersecurity work in that government cloud region.
- OpenAI says Check Point uses Daybreak to speed vulnerability research into customer defenses
OpenAI reports that Check Point is using OpenAI Daybreak so analysts can turn vulnerability intelligence into customer protections faster, framing Daybreak as an accelerator for the research behind new security products.
- OpenAI says Block's security team uses Daybreak to triage every alert and propose patches, turning weeks of work into hours
In an OpenAI customer story, Block, the company behind Square, Cash App and Afterpay, says its security team uses OpenAI's Daybreak and what OpenAI calls the OpenAI Defense Factory for vulnerability discovery, detection and response, with a human validating every result. Block says triage throughput and intelligence gathering are each up 50%, an autonomous hunting system performs about 1,100 analyst-hours of investigation per quarter, and detection changes can be replayed against seven months of telemetry in under an hour. In a red team exercise, the model chained SQL injection, privilege escalation and lateral movement into a full attack path in about two hours, and the time from a high-level vulnerability to an automated proposed patch was 18 minutes, according to Block. Block's harness on top of Daybreak routes tasks to different models and lets humans take over.
- OpenAI says SentinelOne uses OpenAI Daybreak to reverse-engineer malware in about eight hours instead of four to six weeks
In an OpenAI customer story, SentinelOne Chief AI Officer Gregor Stewart says OpenAI Daybreak, which gives defenders access to OpenAI models for cybersecurity work, produces human-level malware reverse-engineering output in roughly eight hours, work that takes a top researcher four to six weeks. Stewart says the speed lets SentinelOne analyze far more samples and find shared components and relationships across threats that isolated investigations missed, and gives more team members access to scarce security expertise, while humans stay accountable for decisions such as architecture changes. SentinelOne plans to bring OpenAI's cyber capabilities into code creation so developers can fix security issues as they build.