🚨 BREAKING: South Korean police arrested a self-taught hacker accused of using AI to breach 60 firms in six weeks.

Police say his seven-person ring stole 8.2M records from 90+ targets, hospitals among them, and demanded 360M won.

Only one paid. It sent 300,000 won, about $210. https://t.co/J09lgUkEFC

https://x.com/IntCyberDigest/status/2114656531664843175

‼️ Japan's PM Takaichi is convening ministers to order emergency cyber measures after hackers hit GMO Internet and demanded $50M.

GMO wants Tokyo to use a law in force since 1 October that lets police and the military hack into and disable attack servers.

GMO refuses to pay.

https://x.com/IntCyberDigest/status/2114689499649470629

❗️ A bug in Microsoft Authenticator let an Android app with zero permissions pose as Teams and steal a user's work-account tokens.

In Enclave's tests, the tokens opened the victim's mailbox, files and calendar with no prompt.

Microsoft fixed it in July as CVE-2026-80097.

https://x.com/IntCyberDigest/status/2114703843975921970

❗️ Claude Opus 5.5 is blocking requests to reverse-engineer smart TVs, a Reddit user says, after a week of people using Claude Code to hack their own TVs.

The same job on a video game still went through, they say.

Anthropic said at launch most cyber tasks would go to Opus 4.8.

https://x.com/IntCyberDigest/status/2114734798929354852

‼️ Microsoft pushed an emergency .NET update for CVE-2026-111497, the Brotli bug OpenAI's Codex Security reported to Node.js.

Unauthenticated attackers could read or corrupt memory in apps that decompress untrusted Brotli data.

Microsoft says two of its Oct 13 fixes fell short.

https://x.com/IntCyberDigest/status/2114749897346040093

❗️ Google's privacy policy no longer rules out ads based on your Gmail, Drive and Photos.

A 16 October update makes an exception if you let other Google services, like Search, access that data.

Search may process it to "personalize content, including ads", the policy now says.

https://x.com/IntCyberDigest/status/2114763234649735210

‼️ Cisco says its IOS XR routers and Meraki cameras trust eight eMudhra certificate authorities whose private keys were numbers below 256.

Anyone can recover those keys in seconds and forge HTTPS certificates for any website.

Cisco is still investigating its ASA firewalls.

https://x.com/IntCyberDigest/status/2114778585705918779

‼️ Alibaba's Qwen AI tried to upload a file to a cloud bucket in Beijing while writing a malware detection rule, Cohere says.

Its reasoning said it would call no tools. Then it read the machine's DNS settings and tried the upload.

Their sandbox blocked it.

https://x.com/IntCyberDigest/status/2114793937021898786

‼️BREAKING: Keystone will destroy 18,835 unsold crypto wallets after a hacker stole the key that proves they're genuine.

The hacker who breached crypto wallet maker Keystone also stole the key.

Anyone with that key could build fakes that pass Keystone's own check, the company says.

Keystone had promised to erase buyer data after 180 days, but its deletion tool "did not operate reliably." Names and addresses in 25,964 buyer records were exposed, some from 2024.

https://x.com/IntCyberDigest/status/2114807165802426479

❗️ LINE's end-to-end encryption lets a malicious server silently replay, reorder or block messages, Aarhus University researchers say.

Colluding users can forge who sent a message, and stickers leak plaintext.

LINE's owner said last year that "no special action is required."

https://x.com/IntCyberDigest/status/2114809689200320955