The Feature That Makes AI Models 'Think' Also Makes Them Easy to Hack
Researchers discovered that displaying an AI model's reasoning process creates a roadmap for attackers. OpenAI's o1 rejection rate dropped from 98% to under 2%.
Tag
#gemini
PromptSpy: First Android Malware Weaponizing Generative AI
ESET discovers Android malware that queries Google's Gemini AI in real-time to navigate infected devices and maintain persistence across any Android version.
AI Privacy Audit: What Eight Popular Assistants Collect
A source-by-source audit of eight AI assistants, what they collect, how training defaults differ, and which privacy settings users can change.
Gemini 3.1 Pro More Than Doubles Its ARC-AGI-2 Score
Google's Gemini 3.1 Pro scores 77.1% on ARC-AGI-2. API pricing starts at $2 per million input tokens for prompts up to 200,000 tokens.
Researchers Prompt-Injected Google Translate's Gemini Mode
Researchers tricked Google Translate's Gemini-based Advanced mode into answering prompts, including requests for drug and malware instructions.
Google's Gemini 3 Deep Think Solves 18 Unsolved Research Problems and Disproves a Decade-Old Conjecture
Google's upgraded reasoning model finds flaws in peer-reviewed papers, optimizes semiconductor fabrication, and outperforms every frontier model on scientific benchmarks.
Perplexity's Model Council Asks Three AIs Instead of One. The Privacy Trade-Off Is Obvious
Perplexity launched Model Council, running your queries through Claude, GPT, and Gemini simultaneously. Multi-model consensus could reduce hallucinations, but it triples your data exposure and costs $200 a month.
Chrome's Auto Browse Lets Google's AI Surf the Web for You - and Watch Everything You Do
Google's new agentic browsing feature streams every page you visit to its servers. Here's what that means for your privacy.
Apple Hands Siri to Google. The Privacy Questions Pile Up.
Apple's deal to power Siri with Google's Gemini raises questions about where your data actually goes -- especially as the two CEOs contradict each other.
Google DeepMind Just Bought an Emotion-Reading AI Company. That Should Worry You.
Three deals in one week -- including a startup that detects emotions from your voice. Google is assembling capabilities that should make privacy advocates nervous.
The Pentagon Gave 3 Million People an AI Chatbot. What Could Go Wrong?
GenAI.mil has 1.1 million users in two months. The military wants Grok next. Between hallucinations, conflicts of interest, and an 'AI-first' strategy that prioritizes speed over safety, the risks are piling up.