International Cyber Digest
6.28K subscribers
924 photos
52 videos
2 files
168 links
Independent reporting on cybersecurity, tech, AI & digital policy. Got a tip? http://internationalcyberdigest.com/tips
Download Telegram
Spotify says it removed 75 million spammy, low-effort tracks in a year and that fully AI-generated music draws under 1% of streams.

Spotify's AI strategy splits into two tracks: purge the industrial slop, and ask honest artists to self-label. AI credits are voluntary and in beta.

Australia is an interesting test case: A Like a Prayer cover sits #1 song on Aussie radio right now with 35 million+ Spotify streams, and after musicians questioned whether a human performed it, producer Josh Fawaz admitted he uses AI as a tool.
❀5😁1
❗️ A lenticular spoof ad at a bus stop near Meta's London headquarters flips TV personality Kylie Jenner's smart glasses campaign into a skeletal surveillance warning depending on where you stand.

The design borrows from They Live, the 1988 film where special sunglasses expose the manipulation hidden inside advertising.

The underlying concern is real, since the latest Ray-Ban Meta generation records up to three minutes of continuous video, has stealth mods and researchers have demonstrated again and again that footage can be matched to names and addresses using facial recognition.
❀15πŸ‘7πŸ’―1
‼️ Linus Torvalds told Linux kernel developers this week that Linux is not one of those anti-AI projects and that AI is now a clearly useful tool.

He says AI's usefulness was still arguable a year ago but no longer is, and that the doubters simply haven't used the tools. Nobody is forced to use AI, but he will ignore anyone trying to stop others from using it. Anyone who wants an anti-AI project, he says, can fork the code or walk away.

In 2024, Torvalds dismissed AI as mostly marketing. By this January he was shipping AI-generated Python in his own hobby repo, and the kernel now accepts AI-written code under disclosure rules.
πŸ‘9😒7πŸ’©4😨3❀1🀣1
❗️ UPDATE: Samsung backed down after a Samsung Health warning appeared to threaten users with health-data deletion and broken sync if they refused AI training.

Samsung now says opting out only deletes data collected for AI development, not existing Samsung Health records.

That clarification came after users noticed the obvious problem... Health data is not AI training bait.
😁4πŸ’©3
❗️ An Irish Government minister dined with Meta lobbyists at the US ambassador's residence in Dublin, as part of a wider US push to soften Europe's policing of big tech.

Ireland just took over the EU Council presidency on July 1 and holds it until the end of this year. It is also home to the regulators that police Meta under EU privacy and platform rules. Meta reported $196 billion in ad revenue last year, says the EU Digital Markets Act is threatening their business model.

Brussels has noticed the conflict of interest. 60 academics are calling on Ireland to recuse itself from digital and tax files, German MEP Andreas Schwab wrote to Irish EU commissioner Michael McGrath citing "serious questions" over neutrality, and TΓ‘naiste Simon Harris is defending Ireland's foreign investment model as a "bridge across the Atlantic".

Whoever runs the presidency helps set the EU agenda for six months. That is exactly the window Meta and the Trump administration want.
πŸ’©12
‼️ BREAKING: A hacker breached AI music company Suno using the Shai-Hulud worm and released source code showing how its training set was built:

- 113,879 hours of YouTube Music
- 62,117 hours of Pond5
- 12,287 hours of Deezer
- plus Genius lyrics and a plan to download roughly 1 million hours of podcasts

The same intrusion reached customer emails, phone numbers, and Stripe payment data for what the hacker says is hundreds of thousands of users.

Suno dates the incident to November 2025, calls it limited, and says individual notifications were not warranted. Customers are surprised to not have been informed about the breach.
πŸ’©16πŸ”₯6❀1
‼️ Tech Against Terrorism tested 27 AI models with 2,000+ prompts seeking bomb-making and weapons guidance. ChatGPT refused only 48% of them, the UN-backed group's new report found.

The report documents 30+ cases of AI supporting extremist attacks, linked to 70+ deaths, and flags open-source models as the harder problem because safeguards can be stripped out entirely.

Australia is already moving. The report lands as a 13-year-old faces extremism charges over alleged AI-generated mass shooting scenarios, and Australian researchers are already pushing for turnover-based fines or geoblocks on noncompliant platforms.
πŸ€”4❀2πŸ’©2🀯1