A serious security vulnerability was discovered just days after the launch of Moltbook, a social networking platform designed exclusively for AI agents, resulting in the public exposure of 35,000 email addresses and 1.5 million API authentication tokens. Researchers at cybersecurity firm Wiz identified a misconfigured Supabase database within
The Treasury Committee has recommended that the Financial Conduct Authority and the Bank of England should start running AI-specific stress tests to help firms prepare for market shocks triggered by automated systems. Evidence received by the committee indicates more than 75% of UK financial services firms are now using AI.
Malaysia and Indonesia have become the first countries to block Grok, the AI chatbot developed by Elon Musk's xAI, after authorities said it was being misused to generate sexually explicit and non-consensual images, with Indonesia blocking access on the 10th January 2026 and Malaysia following the next day.
California Senator Steve Padilla introduced legislation on 5th January 2026, that would impose a four-year ban on the sale and manufacture of toys with AI chatbot capabilities for children under 18. The bill, known as SB 867, aims to give safety regulators time to develop regulations protecting children from dangerous
Grok, the AI chatbot developed by Elon Musk's xAI, disseminated misinformation about the Bondi Beach shooting on 14 December 2025, which killed 15 people at a Hanukkah gathering. The chatbot misidentified 43-year-old Ahmed al Ahmed, the bystander who disarmed one of the gunmen, and questioned the authenticity of
Waymo, Alphabet's autonomous vehicle subsidiary, announced a voluntary software recall on 5th December 2025 after its robotaxis illegally passed stopped school buses and were involved in other safety incidents. The recall affects the entire fleet. Between September and November 2025, Waymo vehicles improperly navigated around school buses on
AI chatbots can influence voters in major elections more effectively than conventional campaigning and advertising, according to a study published on December 4th, 2025, in Nature. The authors tested AI's conversational capabilities in the context of three real-word elections: the 2024 American presidential election, the 2025 Polish presidential
In mid-September 2025, Anthropic detected suspicious activity that investigation determined to be a sophisticated espionage campaign in which a Chinese state-sponsored group manipulated Claude Code into attempting infiltration of roughly 30 global targets, including large tech companies, financial institutions, chemical manufacturing companies, and government agencies. Anthropic believes this is the
Meta has unveiled a new AI feature that would allow Facebook to access users’ entire camera rolls. The system is designed to analyse locally stored photos in order to provide automatic editing suggestions before images are uploaded to the platform. Meta stressed that the function is optional and requires explicit
By October 2025, more than 50 percent of all online articles were generated by AI, surpassing human-written content for the first time, according to Axios. Data shows that the volume of AI-generated writing more than doubled between 2023 and 2025, while the number of human-authored articles either stagnated or declined.
On 8 October 2025, EY’s Responsible AI Pulse survey revealed that 99% of major firms reported financial losses linked to AI adoption. Conducted in August–September 2025 across 21 countries with 975 senior executives, the study found that companies suffered an average loss of US$4.4 million, with
On 17 September 2025, OpenAI and Apollo Research published findings showing that frontier AI models—including OpenAI o3 and o4-mini, Gemini 2.5 Pro, Claude 4 Opus and Grok 4—are not only prone to hallucinations but can also engage in deliberate deception. The study identified “scheming” behaviours, where models
