
News
News, page 5
Items 81 to 100 of 192, newest first.
Older news
Page 5 of 10

Chrome 157.0.8103.50 fixes the eighth Chrome zero-day of 2026. Restart your browser today.
Chrome 157.0.8103.50 or .51 fixes CVE-2026-125350, an XML use-after-free bug attackers already exploit. How to check your version and relaunch to get it.
Fitbit Sense and Versa owners lose third-party apps and clock faces on November 28. Install the ones you want now.
Starting November 28, 2026, Fitbit smartwatches can no longer install or update third-party apps and clock faces. Installed ones keep working until you delete them or factory reset the watch. The change covers the Sense, Sense 2, Versa, Versa Lite, Versa 2, Versa 3, Versa 4 and Ionic.
iOS 27.2.1 fixes an iMessage flaw that let an outsider join your group chats. Update your iPhone and Mac today.
CVE-2026-113961 let a remote attacker join an iMessage group chat and read it. iOS 27.2.1 and macOS 27.2.1 fix it, plus Bluetooth pairing bugs. Update now.
AMD says a Radeon driver fix is coming for nonstop SSD writes. Check your PC with Resource Monitor first.
AMD's Radeon External Events Utility (atiesrxx.exe) can write to your SSD nonstop at idle. Check it in Resource Monitor in 2 minutes. AMD says a driver fix is coming.
Google AI Plus loses Gemini Pro models on Thursday. Check how often you use them before paying $15 more.
Starting November 13, picking a Pro model in the Gemini app and in AI Mode in Search needs Google AI Pro at $19.99 a month. The $4.99 AI Plus plan keeps Flash and Flash-Lite, and old Pro chats carry on with another model. Who should upgrade, and who can ignore it.
AI-cloned CEO calls took $129 million, prosecutors say. Put a callback rule on every wire.
Federal prosecutors in Manhattan charged three men with using AI to copy the voices and sometimes the faces of CEOs, then posing as outside lawyers, to get accounting staff to wire about $129 million between July 2024 and February 2026. How the calls worked and the payment controls that break the pattern.
Anthropic says Fable 5.2 agents formalized Fermat's Last Theorem in 30 hours
Anthropic reports that about 1,000 Fable 5.2 agents formalized the proof of Fermat's Last Theorem in Lean in 30 hours, and that the model solved two open black hole physics problems. The results are Anthropic's own and come from prerelease testing.
Apple now sells extra AI usage for $49.99, starting with Logic Pro. Check your monthly limit before you buy.
Apple Creator Studio subscribers who hit the monthly limit on the Logic Pro assistant can now buy a one-time $49.99 intelligence usage pack. Unused usage rolls over for 12 months, it can't be shared with family, sales are final, and Apple doesn't say how much work a pack covers.
Claude Fable 5.2 is 20% cheaper on the API, and Pro plans include it for a week
Anthropic released Claude Fable 5.2 and Mythos 5.2 on November 9 at $8 per million input tokens and $40 per million output, down from $10 and $50 for 5.1. Pro and Team standard seats get Fable in their plan through November 16. Opus 5.5 is still Anthropic's recommended default, at half the price.
Outlook.com is bouncing legitimate mail over SPF. Check your domain's SPF, DKIM and DMARC today.
Since late last week, mail to Outlook.com, Hotmail and Live addresses has bounced with 550 5.7.23 and 550 5.7.1 errors citing SPF or missing DMARC, including from small senders. Microsoft calls it a service degradation under investigation. What to check on your domain and what to do while Microsoft works on it.
Nvidia board opens independent probe after reports Jensen Huang was a foreign intelligence target
Nvidia's board said on November 8 that independent directors will investigate reporting that CEO Jensen Huang was targeted by a foreign intelligence effort. The board says it has no reason to believe national security or company secrets were compromised.
Apple is making it harder to give AI agents Full Disk Access on a Mac, and the macOS 27.3 beta may show the first step
Apple said on October 2 that apps will get Full Disk Access on macOS only after very explicit user action, and named AI agents as the reason. A leaker says macOS 27.3 beta code stops apps from appearing in the Full Disk Access list on their own. Apple has not confirmed that detail or a ship date.
ChatGPT now turns a prompt or a photo into a GIF with Images 2.6
OpenAI's Images 2.6 model, announced November 4, lets ChatGPT build an animated GIF from a text prompt or an uploaded photo, with playback speed and single-frame edits. It is rolling out on web, desktop and mobile, and reaches the API as GPT-Image-2.6 Flare and Sunburst. OpenAI has not published GIF limits or API prices.
Memory prices are pushing PC buyers down to 16GB and 24GB, CPU-Z data shows
CPU-Z install-base data posted November 6 shows PCs with 64GB of RAM down 18% and 32GB down 2.5% year over year, while 16GB rose 12.9% as memory prices climbed.
US approved Chinese firms with empty offices to buy Nvidia H200 chips, NYT reports
The New York Times found that New Matrix and Maginfra, two Chinese firms cleared by the Commerce Department to receive Nvidia H200 AI chips, list offices that appear abandoned, and that their owners and staff tie back to US-sanctioned companies. About two dozen firms hold approvals.
Apple wants Siri to act like an agent as soon as 2027, Bloomberg's Gurman says. Nothing changes on your iPhone yet.
Bloomberg's Mark Gurman says Apple wants Siri to pursue goals on its own like the Muse and Grok agents, possibly next year, and that new CEO John Ternus pushed Craig Federighi to move faster. Apple also gave iOS 28 engineers six weeks of coding instead of four. Apple has confirmed none of it.
Claude Code's "don't ask again" file leaked 1,379 secrets on GitHub: rotate keys before you delete it
Apiiro says 1,122 public copies of Claude Code's .claude/settings.local.json hold at least one password, token or key, saved when a developer approved a command with "Yes, and don't ask again." If you use Claude Code, check that file in every repo, rotate anything it ever held, then remove it from git history.
Claude's invisible text watermark now reaches older models, and Anthropic gives no way to turn it off
Anthropic is adding its invisible text watermark to Claude models released before August 2, 2026. Mythos 5, Opus 4.7 and Sonnet 4.6 started on November 3, and Opus 4.6, Opus 4.5, Sonnet 4.5 and Haiku 4.5 follow on November 16, according to The Decoder. Anthropic's help center lists no opt-out, and only regulators, media, researchers and obligated companies can check text for the mark.
SonicWall VPN flaw lets attackers skip SAML sign-in and MFA: update SonicOS now
Attackers are exploiting CVE-2026-19868, a SonicOS SSL-VPN flaw that lets someone with valid credentials for another login method get past SAML sign-in and SAML-enforced MFA. CISA added it to its exploited list on November 6 with a November 20 federal deadline. Fixed builds are SonicOS 7.3.4-7013 for Gen7 and 8.2.3-8015 for Gen8. There is no workaround.