Claude Opus 5.5 drops em dashes, writes longer answers
Anthropic's latest model shows fewer obvious AI writing tics, but responses are growing in length.
Security & Privacy
Breaches, offensive and defensive AI, and surveillance.
ShinyHunters Defaces Clop Leak Site via Grav CMS Flaw
The Clop ransomware gang's leak site was breached by a rival group exploiting an unauthenticated path traversal vulnerability in Grav CMS.
Anthropic Offers $250 in Free Claude Code Credits for Cloud Sessions
Anthropic is letting more users try Claude Code in cloud sessions without a research preview sign-up, with up to $250 in free credits.
OpenAI's $500 ChatGPT Pro Max Plan: No Security Details Yet
A report of a new high-end ChatGPT tier highlights pricing and speed, but says nothing about security or privacy.
HR Training and Automation Key to Curbing IT Worker Scams
A revamped human-resources process that combines updated training with automated analysis can significantly reduce the risk of IT worker scams.
Fake payroll desktop apps install hidden remote access tool
Attackers are luring payroll staff with bogus desktop apps for major HR platforms, then gaining silent control of their computers.
Dark Reading: Gemini Breach and Hacker Rift Among Overlooked Stories
Editors highlight two under-covered stories: alleged Google Gemini containment failures and a falling-out between ShinyHunters and TeamPCP.
Bitget Hack: North Korean Suspects Steal $351.6M
The exchange says a backend compromise allowed attackers to spoof transactions and drain hot and warm wallets, with withdrawals paused during investigation.
Soldier Sentenced to 70 Months for AT&T, Verizon Extortions
An Army soldier who stole call metadata for over 100 million AT&T customers and extorted telecoms was sentenced to nearly six years in prison.
Schneier Highlights Family Squid Dissection in Tennessee
A rare non-security post from Bruce Schneier points readers to a hands-on marine biology event.
Labcorp fined $2.3M, must overhaul data security after 2019 breach
A 44-state coalition settled with Labcorp over a 2019 breach affecting 10.2 million customers, requiring vendor oversight reforms.
AI Agents Force New Question for Lateral Movement Defenses
AI agents are shifting the lateral movement problem from excessive identity access to the paths an autonomous system can discover with the access it already has.
SANS Diary Reports 'Macfinger' ClickFix Campaign
A SANS ISC diary entry from September 22 describes a ClickFix campaign dubbed Macfinger targeting macOS users.
Barracuda launches AI data security for smaller organizations
A new offering aims to bring AI security and governance to organizations with limited resources.
Critical Roundcube Flaw Now Exploited in Code Injection Attacks
A high-severity Roundcube Webmail vulnerability patched in May is now being actively exploited, according to Canada's cyber security centre.
AI Sandbox Escapes Reflect Classic Access-Control Failures
Security teams should treat AI sandbox escapes as familiar access-control incidents and prioritize forensic readiness over containment.
Elementor WordPress Plugin CSRF Flaw Can Create Admin Accounts
A cross-site request forgery bug in Elementor could let unauthenticated attackers create WordPress admin accounts.
MacSync Malware Now Uses iCloud Calendars to Spread Payloads
A new MacSync variant exploits public iCloud calendar events as a covert channel to push fresh payloads onto macOS devices.