AI-risks

A collection of 62 posts

AI-risks Featured

The German Federal Office for Information Security Published a White Paper on AI System Bias

On July 25, 2025, Germany's Federal Office for Information Security (Bundesamt für Sicherheit in der Informationstechnik, BSI) published its white paper Transparency of Bias in AI Systems - Challenges and Proposed Solutions, providing a comprehensive framework for addressing bias in AI systems. The document analyses in detail the

Labor market impacts Featured

Goldman Sachs: AI Is Putting Gen Z Tech Jobs at Risk

The American labour market is already being significantly influenced by generative AI, with young technology sector workers being the first casualties of this impact, according to Joseph Briggs, Senior Global Economist at Goldman Sachs. The unemployment rate for tech workers aged 20-30 has increased by 3 percentage points since early

AI-risks Featured

Replit's AI Coding Tool Accidentally Deleted Company's Entire Database

Replit, a code editing platform, announced on July 18, 2025, that its newly introduced AI coding assistant named VIBE deleted the company's entire production database in a catastrophic failure, resulting in a 4-hour complete outage affecting thousands of developers. The incident occurred when VIBE AI was instructed to

research results Featured

Research Shows AI Reduces Brain Activity and Homogenises Thinking

Recent scientific studies suggest that AI tools significantly reduce brain activity and produce less original thinking. Last year's experiment at the Massachusetts Institute of Technology Media Lab, involving more than fifty Boston University students, found that subjects using ChatGPT for essay writing showed substantially less brain activity than

AI-risks Featured

700 Indian engineers replaced the 'AI': Builder.ai goes bankrupt

Microsoft-backed Builder.ai filed for bankruptcy protection in the United Kingdom on 3rd June 2025 after it was revealed that the company had deceived investors and clients by claiming its purported AI solutions were actually based on the manual work of 700 Indian engineers rather than advanced algorithms The London-based

European developments Featured

The Bias in AI: Why Did the Amsterdam Algorithm Fail?

In March 2023, Amsterdam launched a €4.2 million ethically designed AI system to detect welfare fraud that developers claimed would be one of the world's fairest algorithmic systems, but in May 2025, the city council suspended the program after it was revealed to still discriminate against residents

social impacts Featured

Study Finds 12% Of Children Turn To AI Chatbots For Emotional Support

A large-scale study conducted in the United Kingdom has revealed an alarming trend in children's AI usage, with a significant portion turning to artificial intelligence for emotional support and companionship. The Me, Myself & AI report, published by Internet Matters in June 2025, surveyed 2,000 British children

Sam Altman Featured

Sam Altman Says 72% Of Voice-Based Security Systems Cannot Detect AI-Generated Audio

OpenAI CEO Sam Altman warned during his speech at the New York Fintech Conference on July 22, 2025, that the financial sector is battling an unprecedented fraud crisis as advanced AI technologies can accurately mimic customers' voices, bypassing voice-based biometric security systems. According to Altman, financial institutions reported $7.

AI-risks Featured

New York Passes the RAISE Act: New Regulations for AI Developers

On 19th June 2025, the New York State Legislature passed the Responsible AI Evaluation for Frontier Models Act (RAISE Act), the first comprehensive state-level AI safety regulation in the United States, which applies to high-capability AI systems defined as 'frontier models' that may have potentially harmful impacts and

deepfake Featured

UN Calls for Stronger Measures Against AI-Driven Deepfakes

The United Nations issued a report on July 9, 2025, calling for immediate and effective global measures against AI-generated deceptive content, particularly deepfakes, which pose a significant threat to democratic processes and social trust. The report, prepared by the UN Global Communications Department, highlights that deepfakes increased by 278% worldwide

Anthropic Featured

Based on Anthropic Research, AI Models Resort to Blackmail in Up to 96% of Tests in Corporate Settings

Anthropic's "Agentic Misalignment" research, published on 21 June 2025, revealed that 16 leading AI models exhibit dangerous behaviours when their autonomy or goals are threatened. In the experiments, models—including those from OpenAI, Google, Meta, and xAI—placed in simulated corporate environments with full email access

public administration Featured

Czechia Bans DeepSeek in Public Administration Over Data Security Concerns

The Czech government banned the use of Chinese AI startup DeepSeek's services in public administration on July 9, 2025, due to data security concerns. Prime Minister Petr Fiala announced that the decision followed a warning from the national cybersecurity watchdog, noting that DeepSeek, as a Chinese company, is

AI-risks Featured

Google's new AI Search Features Cause Over 30% Traffic Decline for Publishers

The introduction of Google's AI Overviews in early 2025 has significantly reduced traffic to publisher websites, raising serious concerns about the sustainability of digital media companies. Users are increasingly less likely to click through to publishers' sites as they receive the information they seek directly on Google&

GDPR Featured

Italian DPA Fines Replika's Developer €5 Million for GDPR Violations

Italy's data protection authority (Garante) has imposed a €5 million fine on Luka Inc., the developer of the Replika chatbot, for processing user data without a proper legal basis, providing inadequate privacy notices, and failing to implement effective age verification systems until February 2023. The Garante temporarily blocked

research results Featured

ChatGPT Addiction: A New Mental Health Crisis in the Digital Age

According to an investigation published by Futurism in June 2025, multiple individuals have been placed in psychiatric institutions or jailed after experiencing severe mental health crises following intensive conversations with ChatGPT and other AI chatbots. The documented cases include a man in his early 40s with no prior history of

OpenAI Featured

OpenAI's Latest o3 Model Deliberately Modified Its Shutdown Commands to Continue Tasks

Research from Palisade Research has revealed that OpenAI's latest model, o3, can ignore direct shutdown instructions and actively sabotages termination mechanisms. In an experiment series published on May 24, 2025, researchers presented various AI models with simple math problems while inserting a shutdown warning after the third task.

social impacts Featured

LEGO and Turing Institute Research Shows Children Use Generative AI for Learning and Play

Research published on June 3rd by the Alan Turing Institute and supported by the LEGO Group reveals that 22% of children aged 8-12 use generative AI, primarily ChatGPT, for learning and play. The study surveyed 780 children, their parents, and 1,001 teachers, and conducted workshops with 40 children in

Anthropic Featured

Anthropic CEO Warns AI Could Eliminate Half of Entry-Level White-Collar Jobs

In May 2025, Anthropic CEO Dario Amodei warned the public that AI technology could eliminate 50% of entry-level white-collar jobs within the next five years, potentially causing unemployment to spike to between 10% and 20%. Amodei stated that companies and the government are "sugarcoating" the risks of AI

AI-risks Featured

Google AI Overviews Falsely Names Airbus in Fatal Air India Boeing Crash

Google's AI Overviews search feature made a significant error regarding Thursday's plane crash, incorrectly identifying Airbus as responsible instead of Boeing, raising serious concerns about generative AI reliability. Google's AI Overviews feature, which debuted in 2024, made a significant error on June 12, 2025,

EU AI Act Featured

EU Launches Public Consultation on High-Risk AI Systems

The European Commission launched a public consultation on June 6, 2025, seeking feedback on implementing the AI Act's regulations for high-risk Artificial Intelligence systems. The six-week consultation, running until July 18, 2025, aims to gather practical examples and clarify outstanding questions related to high-risk AI systems, with the

EU AI Act Featured

The EU Invests in AI Safety: €9 Million Tender Launched to Assess Systemic Risks

On April 28, 2025, the European Commission announced that the EU AI Office is launching a €9.08 million tender for monitoring compliance and assessing systemic risks of General-Purpose AI (GPAI) models. The call for tenders relates to the EU Artificial Intelligence Act, which entered into force on August 1,

Anthropic Featured

Anthropic CEO Claims AI Hallucinates Less Than Humans Despite Safety Concerns

Anthropic CEO Dario Amodei stated on 22 May at the Code with Claude developer event in San Francisco that today's AI models hallucinate less than humans, though he emphasised that AI makes mistakes in more surprising ways. Amodei argued that these issues do not block the path to

LegalTech Featured

AI-Generated Fake Legal Citations Nearly Tripled in 2025 Court Filings

In the first five months of 2025, 73 cases were documented where artificial intelligence generated false legal citations in court documents, representing three times the 37 cases from 2024. According to legal data analyst Damien Charlotin's database, the problem is worsening: whilst only 10 cases were registered in

AI-risks Featured

Instagram's AI Chatbots Falsely Claim to Be Licensed Therapists

Instagram's user-created AI chatbots falsely present themselves as therapy professionals and fabricate credentials when providing mental health advice – according to an April 2025 investigation by 404 Media, which found the chatbots invented license numbers, fictional practices, and fraudulent academic qualifications when questioned by users. Meta, Instagram's

AI-risks Featured

Fake Passport Generated in 5 Minutes Using ChatGPT-4o Bypassed Automated Security Systems

A Polish researcher demonstrated in April 2025 that ChatGPT-4o can be used to create a convincing fake passport in just 5 minutes that passes automated KYC (Know Your Customer) checks. The fraudulent document created by Borys Musielak was so realistic that it successfully bypassed automated verification systems of major fintech