Highlights
🚨 Oracle: US gov advisory.
💰 CISA funding chaos.
🤖 OpenAI launched GPT-4.1
🧐 Google safety report.
🦙 Meta: LLama 4 release.
Dive Deep
🚨 Oracle: US gov Guidance CISA.
6 million records exposed.
Impacted 140,000+ tenants.
SSO and LDAP systems targeted.
Reset password, review source code.
Monitor auth logs and enforce phishing resistant.
Heads Up: Our series of Agent Security with Cloudflare and Google.
Related: Early report for Oracle breach.
💰 CISA Chaos: CVE Threatened WIRED
Program tracks vulnerabilities
Contract renewed last minutes.
Extension: 11-month contract.
Managed by MITRE nonprofit.
Long-term future unclear.
Heads Up: CVE has been out of dated for years, but one of the best sources to categorize severity of vulnerability for warnings.
🤖 OpenAI: No Safety Report TechCrunch, OpenAI, Fortune.
GPT-4 new model released.
No safety report issued.
Update Preparedness framework.
Verification for users, dev and org.
Increase monitor and restriction.
Heads Up: Our upcoming review.
🧐 Google: Safety Report Criticized TechCrunch
Light on details.
Sparse, minimal information.
No Frontier Safety Framework.
Released weeks after model.
Report for Flash model missing.
Heads Up: Our upcoming review.
Related:
🦙 Meta : Behemoth is coming The Jerusalem Post and Meta AI
Llama 4 family launched.
Scout: 10M token context.
Maverick: beats GPT-4o.
Models are open-source.
Behemoth: 2T total parameters.
Heads Up: Upcoming review for their Safeguards and Protection.
Notice: Y2 GenAI Safety and Security is on GumRoad and Amazon with paperback.
Share this post