A new TamperBench study found leading open-weight AI models could all be tampered with despite existing safeguards.

Statista Consumer Insights surveyed 32 countries, finding varying concern about AI-generated news among respondents.

Personalized AI may support different personality tendencies, but research suggests it can also reinforce or influence users’ ...

AI chatbots simplify information seeking but may increase trust while reducing source-checking and exposure to disagreement.

US GenAI chatbot growth is slowing, but Meta AI expands its share while competitors experience mixed August results.

At least one-third of Americans consider price important when purchasing every surveyed consumer technology category.

Researchers warn poorly designed spy detectors may create dangerous false security, especially for domestic abuse survivors ...

Texas officials increasingly scrutinize ordinary Muslim religious practices, portraying them as political or security threats ...

AI summaries reduce website traffic, weakening publishers’ incentives to create reliable content and support the open web.

Surfshark analyzed 40 popular apps across four categories, finding browsing-history collection widespread across mobile platforms.

Google’s AI summaries generally debunked conspiracies, but explanations were often too cursory to effectively counter ...

Cyberbullying may increase teenagers’ later emotional and behavioural problems, especially through loneliness and difficulty ...