News

De gemeente Amsterdam moet klachten over ongewenst gedrag beter afhandelen, vindt de Ombudsman Metropool Amsterdam.

Правительство префектуры Гифу в Японии заявило, что начнет использовать беспилотники, издающие звуки лая собак, а также звуки фейерверков, чтобы попытаться отпугнуть медведей от жилых районов.

Поставки шведских истребителей Saab JAS 39 Gripen могут стать одним из наиболее сбалансированных и реалистичных вариантов для усиления боеспособности украинских Воздушных сил в среднесрочной перспективе. Западные аналитики сходятся во мнении, что эти машины идеально подходят для украинских условий и могут быть быстро интегрированы в состав ВСУ.

Researchers found AI is hopeless at most Upwork task, it gets the news wrong half the time — and humans crush AI on world model tests. AI Eye. AI agents cant complete 97% of tasks on Upwork to even a basic standard. Researchers at Scale AI and the Center for AI Safety got six different AI models to attempt 240 Upwork projects across categories, including writing, design and data analysis and then compared the results to the real freelancer. The overwhelming majority of the time, the AI models were unable to complete the tasks successfully, with the best AI model, Manus, completing just 2.5% of tasks and earning $1,810 out of $143,991 on offer. Claude Sonnet and Grok 4 managed to finish 2.1% of the tasks. While AI agents are good at simple and defined tasks like “generate a logo,” the research found they are bad at multi-step workflows, taking any initiative or using judgment. So they wont be causing mass unemployment for a while yet. This backs up research from August at MIT, which found that 95% of organizat...

Researchers found AI is hopeless at most Upwork task, it gets the news wrong half the time — and humans crush AI on world model tests. AI Eye. AI agents cant complete 97% of tasks on Upwork to even a basic standard. Researchers at Scale AI and the Center for AI Safety got six different AI models to attempt 240 Upwork projects across categories, including writing, design and data analysis and then compared the results to the real freelancer. The overwhelming majority of the time, the AI models were unable to complete the tasks successfully, with the best AI model, Manus, completing just 2.5% of tasks and earning $1,810 out of $143,991 on offer. Claude Sonnet and Grok 4 managed to finish 2.1% of the tasks. While AI agents are good at simple and defined tasks like “generate a logo,” the research found they are bad at multi-step workflows, taking any initiative or using judgment. So they wont be causing mass unemployment for a while yet. This backs up research from August at MIT, which found that 95% of organizat...

Researchers found AI is hopeless at most Upwork task, it gets the news wrong half the time — and humans crush AI on world model tests. AI Eye. AI agents cant complete 97% of tasks on Upwork to even a basic standard. Researchers at Scale AI and the Center for AI Safety got six different AI models to attempt 240 Upwork projects across categories, including writing, design and data analysis and then compared the results to the real freelancer. The overwhelming majority of the time, the AI models were unable to complete the tasks successfully, with the best AI model, Manus, completing just 2.5% of tasks and earning $1,810 out of $143,991 on offer. Claude Sonnet and Grok 4 managed to finish 2.1% of the tasks. While AI agents are good at simple and defined tasks like “generate a logo,” the research found they are bad at multi-step workflows, taking any initiative or using judgment. So they wont be causing mass unemployment for a while yet. This backs up research from August at MIT, which found that 95% of organizat...

Researchers found AI is hopeless at most Upwork task, it gets the news wrong half the time — and humans crush AI on world model tests. AI Eye. AI agents cant complete 97% of tasks on Upwork to even a basic standard. Researchers at Scale AI and the Center for AI Safety got six different AI models to attempt 240 Upwork projects across categories, including writing, design and data analysis and then compared the results to the real freelancer. The overwhelming majority of the time, the AI models were unable to complete the tasks successfully, with the best AI model, Manus, completing just 2.5% of tasks and earning $1,810 out of $143,991 on offer. Claude Sonnet and Grok 4 managed to finish 2.1% of the tasks. While AI agents are good at simple and defined tasks like “generate a logo,” the research found they are bad at multi-step workflows, taking any initiative or using judgment. So they wont be causing mass unemployment for a while yet. This backs up research from August at MIT, which found that 95% of organizat...

Researchers found AI is hopeless at most Upwork task, it gets the news wrong half the time — and humans crush AI on world model tests. AI Eye. AI agents cant complete 97% of tasks on Upwork to even a basic standard. Researchers at Scale AI and the Center for AI Safety got six different AI models to attempt 240 Upwork projects across categories, including writing, design and data analysis and then compared the results to the real freelancer. The overwhelming majority of the time, the AI models were unable to complete the tasks successfully, with the best AI model, Manus, completing just 2.5% of tasks and earning $1,810 out of $143,991 on offer. Claude Sonnet and Grok 4 managed to finish 2.1% of the tasks. While AI agents are good at simple and defined tasks like “generate a logo,” the research found they are bad at multi-step workflows, taking any initiative or using judgment. So they wont be causing mass unemployment for a while yet. This backs up research from August at MIT, which found that 95% of organizat...

Researchers found AI is hopeless at most Upwork task, it gets the news wrong half the time — and humans crush AI on world model tests. AI Eye. AI agents cant complete 97% of tasks on Upwork to even a basic standard. Researchers at Scale AI and the Center for AI Safety got six different AI models to attempt 240 Upwork projects across categories, including writing, design and data analysis and then compared the results to the real freelancer. The overwhelming majority of the time, the AI models were unable to complete the tasks successfully, with the best AI model, Manus, completing just 2.5% of tasks and earning $1,810 out of $143,991 on offer. Claude Sonnet and Grok 4 managed to finish 2.1% of the tasks. While AI agents are good at simple and defined tasks like “generate a logo,” the research found they are bad at multi-step workflows, taking any initiative or using judgment. So they wont be causing mass unemployment for a while yet. This backs up research from August at MIT, which found that 95% of organizat...

While New Yorkers rallied around an idealist unafraid to be called radical, Israel's opposition leaders shrink from saying what they stand for, beyond replacing Netanyahu

NEAR Protocol (NEAR) joined Internet Computer (ICP) as a top performer, rising 3.3%.

“Vibe coding”, a word that essentially means using artificial intelligence (AI) to tell a machine what you want instead of coding it yourself, was on Thursday named the Collins Word of the Year 2025. Coined by OpenAI co-founder Andrej Karpathy, the word refers to “an emerging software development that turns natural language into computer code using AI”, according to Collins Dictionary. “It’s programming by vibes, not variables,” said Collins. “While tech experts debate whether it’s revolutionary or reckless, the term has resonated far beyond Silicon Valley, speaking to a broader cultural shift towards AI-assisted everything in everyday life,” it added. Lexicographers at Collins Dictionary monitor the 24 billion-word Collins Corpus, which draws from a range of media sources, including social media, to create the annual list of new and notable words that reflect our ever-evolving language. The 2025 shortlist highlights a range of words that have emerged in the past year to pithily reflect the changing world aro...

Amending the structure and the rules of procedure of the Security Council is a key element of UN reform to make the Organization more representative and balanced. Under certain conditions, expanding the number of non-permanent members of the Security Council while simultaneously limiting the veto power of its permanent members is a realistic approach.

De bank Shawbrook en tonijn-in-blikverkoper Princes om maar enkele te noemen, zorgen voor vers bloed op de Londense beurs. Dat was alweer veel te lang geleden. ‘Het is echt een verschil van dag en nacht.’

Het verlangen naar wetenschappelijke onderbouwing in de politiek is diep geworteld in onze maatschappij. Maar met cijfers en feiten bestuur je nog geen land. Daarvoor zijn visie, doelen en tolerantie nodig.

Though Ethereum is still the preferred platform among institutions for asset tokenization, DeFi apps and stablecoin creation, it faces threats that will erode its edge if it doesn't move to meet the market, argues Axelar co-founder and CEO Sergey Gorbunov.

The integration, powered by Chainlink’s NAVLink oracle technology, represents another leap in bridging traditional finance and decentralized finance together.

Bitcoin.com and Concordium have teamed up to introduce age-verified stablecoin payments to 75 million wallets, blending privacy with new compliance standards. Crypto media and wallet platform Bitcoin.com has partnered with Concordium, a privacy-focused layer-1 blockchain, to enable age-verified stablecoin payments across more than 75 million wallets on Bitcoin.com’s network. Announced on Thursday, the integration allows wallet users to verify specific identity attributes, such as age or jurisdiction, without revealing personal details. Verification occurs off-chain through independent third-party providers, and no personal data is stored on the blockchain. Each transaction utilizes zero-knowledge proof technology to verify compliance requirements while maintaining user privacy. Read more

RedStone expands beyond price oracles with Credora, integrating risk analytics across DeFi protocols Morpho and Spark. Modular oracle network RedStone launched Credora, a decentralized finance (DeFi)-native risk ratings platform aiming to bring transparency and credit analytics to lending protocols.  RedStone said on Thursday that it had expanded beyond price feeds into the broader domain of credit, collateral and risk intelligence through its Credora acquisition in September.  At launch, Credora by RedStone integrates with DeFi lending markets Morpho and Sparks to offer dynamic risk scores and default-probability analytics, accessible through an API.  Read more