Latest cybersecurity reporting from selected sources.

Turns out decades-old email tricks still work against some LLM-powered email filters

Command injection vulns land on exploited list after researchers spot abuse attempts

A specific multi-touch gesture bypasses an authentication prompt, allowing anyone to send messages

Adapting existing local LLM project for security and sovereignty purposes and hopes to one day match Mythos

Data purges deemed an example of 'misaligned behavior' that upstart is working to avoid

Models demand trust without offering verification

Newly documented stealer ClickLock comes for the more trusting Mac user with spot of social engineering

Sentencing bookends the biggest cybercrime conviction in UK history

One in six machines still run the old OS as migration stalls and patch deadlines creep closer

t.me borked for a day until platform proved it had no ties to service favored by cybercriminals

Using the admin password, you could be anyone and see anything

It’s possible to leak PII describing 5.7 million people without breaching privacy rules

The Colonel stops taking online orders and may close stores after logistics partner’s systems go down

Three bugs are under active attack, and two more critical holes could add to the pain

Mega hardware vendor reports problems - but Windows maker isn't yet naming affected models

Experts say it’s a useful post-compromise tool, for those with the brain cells required to put it together

Remember when last month's 206 CVEs seemed eye-watering? Yeah, those were the days

Callum Dare encouraged others to carry out dangerous hoaxes, made mini-movies from the footage

Researcher confirms the uploads have stopped, but says xAI's privacy command was not what fixed them

Human did 10% of the job, AI did 90%