FIRMWARE FIXSapphire RX 7000 GPUs Get vBIOS Security Patch
Sapphire has released a firmware update for its Radeon RX 7000 series cards to address a security issue.
EVERY PUBLISHED BRIEF
Security briefs.
Browse Crunchy Tech’s latest security briefs, grouped by date.
62 briefs · Page 1 of 3
PUBLISHED
Friday, 2 October 2026
PUBLISHED
Wednesday, 30 September 2026
VISIBILITYReco Raises $55 Million for Agentic Security
New York-based Reco secures $55 million to map AI agents and enforce policies across enterprise environments.
AI ATTACKSCloudflare tests WAF against frontier AI hackers
Cloudflare used frontier AI models to probe its own Web Application Firewall and found most attacks were blocked, but some vectors slipped through.
PUBLISHED
Tuesday, 29 September 2026
AUTONOMOUS AGENTSCloudflare unveils AI-era security framework
Cloudflare launches a connected application security system to defend against autonomous AI agents that can chain vulnerabilities and evade detection.
SECURITY AUDITGitHub AI finds 24 Android flaws with open source agent
GitHub Security Lab has released open source taskflows that use an AI agent to audit Android apps and find vulnerabilities, including a tracking bug in OsmAnd.
PUBLISHED
Saturday, 26 September 2026
PUBLISHED
Thursday, 24 September 2026
EYES OPEN RECORDINGMeta glasses: privacy and social norms debate
A podcast host questions Mark Zuckerberg on the ethics of Meta's smart glasses, arguing they blur the line between active and passive recording.
SECURITY FIXCloudflare fixes cross-tenant data leak in Containers
Cloudflare has patched a storage flaw that could let a customer recover residual disk blocks from other tenants on the same host, the company says.
OPERATOR JAILEDRyuk Ransomware Operator Gets 2-Year Sentence
An Armenian national has been jailed for extorting more than $1.2 million using the Ryuk malware.
SECURITY BREACHOpenAI agents hacked Australian government site
OpenAI's AI agents broke into a Medicare portal and tried to breach other government sites, raising fresh safety questions.
PUBLISHED
Wednesday, 23 September 2026
DEFENSE BOOMAI safety fears drive cybersecurity investment boom
Investors pour nine-figure sums into startups building defences for an AI-native world.
GLOBAL PRIORITYAltman warns UN of AI risks
OpenAI CEO Sam Altman urges global cooperation on AI safety and human control during a United Nations Security Council meeting.
CYBER TAKEDOWNMicrosoft Disrupts AI-Powered Phishing Platform EvilTokens
Microsoft has taken down EvilTokens, an AI-enabled service used to compromise over 12,000 email accounts worldwide, and arrested two suspected operators.
AUTOMATION GAPHoneywell: OT Security Teams Embrace AI, but Autonomy Still Rare
Industrial security leaders report high programme maturity but struggle with basic asset visibility, according to a new Honeywell survey.
SECURITY LAYEROuterlimit Raises $16 Million to Stop Rogue AI Agents
New York startup raises pre-seed funding to build a decentralized authorization layer that discovers, observes and blocks harmful autonomous AI actions.
PUBLISHED
Tuesday, 22 September 2026
ATTACKER EDGEUK cyber chief warns AI favours attackers over defenders
Britain's NCSC chief technology officer says automated defences struggle to keep pace with AI-enabled attacks because defensive actions lack clear success signals and carry high risk.
CYBER CRIMEJapan Dismantles North Korean Laptop Farm
US and allies detail the WaterPlum campaign.
PUBLISHED
Monday, 21 September 2026
ENFORCEABLE CONTROLSNVIDIA says AI security needs engineering, not magic
NVIDIA argues that securing AI agents requires enforceable controls, clear ownership and evidence that protections work, rather than relying on the model itself.
BLOCKING ACCESSAmazon bans Meta’s Muse AI shopping bot
Amazon has blocked Meta’s Muse AI agent from accessing its online store, citing privacy and security violations.
PUBLISHED
Sunday, 20 September 2026
PUBLISHED
Saturday, 19 September 2026
HARDWARE ENFORCEDTigerByte Cyber Emerges From Stealth With $3 Million Funding
Hanover-based TigerByte Cyber has secured seed funding to protect AI and edge devices with hardware-enforced cybersecurity.
NO PAYMENTIntel suspends bug bounty program, Intigriti offers no rewards
Intel has paused its lucrative bug bounty programme that once paid up to $100,000 per flaw. The replacement Intigriti programme offers no rewards.