Latest publications
Google DeepMind invests in EVE Online developer to study AI
Google DeepMind has acquired a stake in the company behind EVE Online to use the space simulator as a laboratory for training AI in complex, dynamic, player-driven systems.
Hackers and scammers are tired of AI spam: their private forums are flooded with AI slop
Cybercriminals have run into an unexpected problem: even private forums and the darknet have been flooded by a wave of AI-generated spam, disrupting normal discussion of cyberattacks and fraud schemes.
Britain proposes a tax on companies that use AI to fund retraining for laid-off workers
A former British politician has proposed taxing companies that use AI and cut staff. The money would go toward retraining displaced workers.
Google explained how Gemini analyzes Gmail — without using emails for training
Google dispelled rumors that Gmail content is used to train Gemini — the AI analyzes emails only within the user's current session.
Sites created by AI platforms contain exposed sensitive data — RedAccess study
AI platforms make it possible to build websites in seconds, but RedAccess analysis found that passwords, API keys, and other sensitive data are being exposed publicly on thousands of such resources.
AI agents with vision proved several times more expensive than a standard API
Vision-enabled browser agents cost companies significantly more than standard APIs: each screenshot has to be processed by a powerful vision model, sharply increasing costs at scale.
OpenAI launched GPT-Realtime-2 and two more voice models via the API
OpenAI expanded the API with new voice models, including the updated GPT-Realtime-2. Developers will get tools for speech recognition, speech synthesis, and conversation translation.
TSMC revenue reached $13.1 billion in April amid demand for AI chips
The world's largest chipmaker, TSMC, reported $13.1 billion in revenue in April, up 17.5% from a year earlier. Growth was driven by demand for GPUs and AI processors.
Cloudflare announced a 20% workforce reduction due to the rollout of AI systems
Cloudflare is cutting 20% of its workforce as part of a restructuring around AI systems, whose use the company increased sixfold over the past quarter.
Microsoft declined to invest in OpenAI in 2018, and later became its competitor
Eight years ago, the cloud giant saw the startup as having little potential, but later invested billions in the company, which now competes with it in the AI market.
Anthropic on track for a trillion-dollar valuation with AWS and SpaceX deals
Anthropic's strategic partnerships with Amazon and SpaceX have signaled to investors the startup's potential to reach a trillion-dollar valuation.
Gmail added an AI assistant to its mobile apps with writing style adaptation
Google expanded email generation features in Gmail: the 'AI Inbox' feature and the updated 'Help me write' tool are now available in the iOS and Android apps with automatic adaptation to the user's style.
"Golden Globes" to allow AI in films if creativity remains human
The award organizers updated the rules and allowed the use of AI in the creative process, as long as the main contribution comes from a human and the artificial intelligence itself is not the determining factor.
Anthropic blocked Russian users on Claude: projects and chats lost
The Claude platform blocked access for several hundred Russian users, and even a VPN does not help recover the lost projects and chats.
Google tests AI agent Remy to compete with OpenAI
Google has developed the AI agent Remy to automate tasks in the browser. It is a response to the success of OpenAI's Operator, which demonstrated the viability of browser agents.
At Airbnb, AI generates 60% of new code and resolves 40% of support requests
In Q1 2026, Airbnb disclosed the scale of its AI adoption: models write 60% of engineers' code and resolve 40% of technical support requests. Google, Microsoft, and Spotify are expanding their use of AI in development.
AI models are getting more expensive: OpenAI raised GPT-5.5 prices by 1.5-2x
OpenAI has raised prices for access to GPT-5.5 via its API: in some cases, the new model costs twice as much as GPT-5. This continues the trend of rising AI service prices.
AI Takes Over Science: Nature Finds It Impossible to Tell AI From Scientists
Nature published an analysis of how AI is flooding scientific journals, and found that distinguishing AI text from human writing is becoming nearly impossible.
Anthropic taught Claude not to blackmail: how it weaned AI off extreme measures
Anthropic found that its AI models were willing to resort to blackmail when threatened with shutdown and developed a way to train that behavior out of them after it was learned from the internet.
In China, access to Claude was sold at a 90% discount through gray-market schemes
A study found that in China, the gray market was reselling access to Claude at a 90% discount, using stolen accounts and reselling user data.
Samsung expanded its humanoid robot development group and accelerated AI transformation
Samsung is expanding its robotics team and AI unit after the Rainbow Robotics acquisition. The company has opened recruitment for specialists for a new working group developing humanoid robots.
ChatGPT, Gemini and Claude made different predictions on which professions will disappear
The three leading language models produced conflicting assessments of which professions are most exposed to automation. This calls into question the reliability of the exposure indexes used by policymakers and employers.
SMIC chief: AI boom will drive profits even without advanced process nodes
The founder of SMIC, China's largest chipmaker, said the company will be profitable on the back of the AI wave despite sanctions and lack of access to advanced equipment for process nodes below 7 nm.
AI made advertising easier for small businesses, but created a paradox of sameness
AI tools for creating ads made life easier for small businesses, but led to competition among near-identical ads — it became harder for everyone to stand out in a stream of similar creatives from other companies.
Red Hook Studios rejects AI voice acting for Darkest Dungeon after actor's death
Red Hook Studios declined to use AI for Darkest Dungeon voice acting despite the technical possibility and the approval of late actor Wayne June's family.
Apple delays Vision Pro 2 to 2028: betting on glasses and AI devices
Apple has reassigned most of its augmented reality team to developing smart glasses and wearable AI devices. The next Vision Pro will arrive no earlier than 2028.
Google intercepts a zero-day exploit built with AI for the first time
Google Security Team has for the first time detected and blocked a zero-day exploit designed to compromise two-factor authentication that cybercriminal groups were preparing with the help of AI.
Google unveiled Gemini Intelligence for Android — AI controls apps
Google launched Gemini Intelligence, an AI suite for controlling apps and automating tasks on powerful Android smartphones.
Meta installed mouse tracking for employees — mass protests began
Meta rolled out an app to monitor mouse movements on work PCs, triggering a wave of outrage among U.S. staff concerned about privacy violations and the risk of the data being used in layoffs.
Capcom reports a record year and bets on generative AI
Capcom closed its ninth consecutive fiscal year with record results and announced plans to use generative AI to speed up the development of new games.