FCC Turns Emergency Alert Cybersecurity Into a Baseline Requirement
The FCC adopted targeted cybersecurity rules for Emergency Alert System participants after years of warnings about default passwords, unpatched gear, and internet-exposed alerting equipment. Broadcasters and cable operators will have 60 days after Federal Register publication to meet the new baseline.
Apple’s Mac and iPad Price Hikes Show AI Costs Reaching Consumer Devices
Apple raised prices on several Macs and iPads after AI data-center demand pushed up memory and storage costs. The move makes the AI infrastructure boom visible in ordinary consumer hardware prices.
RAISE US Launches $500M AI Workforce Push With OpenAI and Anthropic Backing
RAISE US, a bipartisan nonprofit led by Gina Raimondo and Eric Holcomb, launched with more than $500 million to test AI workforce programs with states, employers, and major tech backers including OpenAI, Anthropic, Microsoft, and Amazon.
Anthropic’s Mythos Test Shows Why AI Cyber Defense Is Becoming Classified Work
An Anthropic Mythos test with U.S. intelligence agencies reportedly found vulnerabilities in highly sensitive government systems within hours. The episode sharpens the policy problem around frontier AI: the same models that can help defenders fix critical software can also compress the timeline for attackers.
OpenAI Launches GPT-5.6 Sol Under Government-Restricted Preview
OpenAI has launched GPT-5.6 Sol, Terra, and Luna in a restricted preview after U.S. government review. The release brings new pricing, API and Codex access limits, stronger cyber safeguards, and a clearer look at how frontier model launches are becoming governed deployments.
curl 8.21.0 Fixes 25-Year-Old libcurl mTLS Bug
curl 8.21.0 fixes 18 security flaws, including CVE-2026-8932, a 25-year-old libcurl mTLS connection-reuse bug. The practical risk is in applications that embed libcurl and change client certificate settings while reusing connection pools.
Citizen Lab Says Russia Used Cellebrite on Activist’s iPhone After Cutoff
Citizen Lab says Russian authorities used Cellebrite forensic tools on activist Andrey Pivovarov’s iPhone months after Cellebrite said it had stopped selling to Russia and Belarus. The case turns phone forensics into a control problem: what happens when extraction tools keep working after a vendor cuts off a customer?
OpenAI’s Codex Data Shows AI Agents Are Becoming Workflow Systems
OpenAI’s new Codex research shows AI agents moving from coding assistants toward workflow systems that run long, parallel, delegated tasks. The strongest signal is not just developer use, but rapid growth among non-developers and organizational users.
Facebook Creator Studio Returns as an AI App for Creator Workflows
Meta is bringing back Facebook Creator Studio as a standalone AI companion app for creators. The test turns audience analytics, comment triage, and posting advice into a conversational workflow instead of another dashboard.
Mandiant Details Cisco SD-WAN Attack That Turned a Malicious CSV Into Root Access
Mandiant says an attacker used rogue Cisco Catalyst SD-WAN peering, admin password manipulation, and CVE-2026-20245 to gain root access through a malicious CSV upload. The new details make the June SD-WAN advisories an incident-response problem, not just a patching task.