Urgent.News

the world's headlines, one feed

AI

Google’s AI Search seemingly leaked unreleased game content and no one knows how

Google has denied that it scans private Workspace Docs to train its AI.

Google’s AI Search seemingly leaked unreleased game content and no one knows how

A solo game developer has come forward alleging that Google’s AI Search unintentionally exposed details about an unreleased character for their tower defense game, Operation Octo. The developer, who goes by Klub Kofta Studio, stated that the AI provided "surprisingly obscure info" during a casual Q&A session about the game. The AI reportedly mentioned the character name "Vantage Tripod."

The developer confirmed that they had never publicly disclosed the character's existence and stored the information in a private Google Docs file. Google responded to the incident by denying any scanning of private Workspace content for AI model training. Over the years, Google’s AI Search has faced criticism for providing inaccurate information, with some instances causing significant public backlash.

Despite this, the AI has also been known to provide surprising accurate details, as seen in this case where it seemingly disclosed confidential information about a small indie game.

Written by urgent.news from Android Authority's reporting — not their text. Machine-written; read the original for the full account.

This story

This is one outlet's version. Read the fullest account.

Read the original at androidauthority.com →

More in AI

Reddit decides who AI recommends: what I found measuring 241 software categories

Buyers have started asking ChatGPT and Gemini which software to use instead of opening ten tabs. So over the last few weeks I ran a simple experiment 241 times: for each software category, I asked the…

  • Reddit appeared in 93% of 241 software categories studied.
  • 1,202 individual answers were provided by Reddit in the study.
  • AI assistants frequently cited vendors' own comparison pages.

Crypto’s infrastructure era arrives, with AI agents poised to reshape demand

Crypto companies are trying to build a second growth engine by turning agents into the next wave of users.

  • AI agents integrated into crypto wallets and payment networks to serve as users.
  • Kraken overhauls app to enable AI agents for real-time market monitoring and trading.
  • Circle expands Arc blockchain for agentic economy with AI agents handling operational tasks.

My local AI ran on the CPU for a week and I didn't notice

My first Ollama setup felt fast. One evening, one model, one question, a clean answer streaming back at a speed that made cloud AI feel unnecessary.

  • Local AI setup ran on CPU for a week unnoticed
  • Oversized model and context length caused GPU VRAM limit
  • Two mistakes: partial offloading and GPU passthrough conflict