Autor Cointelegraph By Stephen Katte

US-based AI firm Anthropic warns AI development is advancing at a pace that could soon see agents building, training and improving themselves without human input — recommending a slowdown in development.In a blog post published Thursday, Marina Favaro, lead at the Anthropic Institute, and Anthropic co-founder Jack Clark said agents can already run code themselves, delegate hours of work to other agents and could be on the cusp of taking over completely.“For most of AI’s history, humans drove every step in its development cycle. But at Anthropic, we are delegating a growing share of AI development to AI systems themselves, which is speeding up our work,” they said.“Taken far enough, and given enough compute, that trend points to an AI system capable of fully autonomously designing and developing its own successor,” Favaro and Clark added. AI development is advancing at a pace that could lead to agents improving without human input. Source: AnthropicThere are concerns over what could happen if AI is able to become smarter on its own. In December, OpenAI said it is researching how to safely develop and deploy increasingly capable AI, including AI capable of recursive self-improvement. “We want these systems to consistently follow human intent in complex, real-world scenarios and adversarial conditions, avoid catastrophic behavior, and remain controllable, auditable, and aligned with human values,” it said. The company is also hiring a researcher for recursive self-improvement preparedness, which forms part of its Safety Research team. AI model improvement has been roughly doubling every four months, rather than every seven months, according to Favaro and Clark. The role of humans is narrowing at each step, with Anthropic’s Claude model authoring around 80% of the code merged into Anthropic’s codebase.“We are not there yet, and recursive self-improvement is not inevitable. But it could come sooner than most institutions are prepared for,” they said. “Once human- and AI-authored code quality reach parity, humans will stop writing code entirely and shift to only reviewing it. But if they can’t review code as quickly as Claude can generate it, human review will become the bottleneck to AI development,” they added.Favaro and Clark also said that slowing development to allow more time to address its “immense” implications would be ideal.Related: Modern robots impress, but are years away from replacing humansIn April, Anthropic ruled out releasing its AI model, Claude Mythos, to the public over concerns about the threat to global cybersecurity. Claude Mythos was able to easily create software exploits, leading Anthropic to rule out a public release for now. Source: AnthropicAt the same time, a group of tech leaders, including some from Anthropic and OpenAI, released an open letter on Thursday, urging lawmakers to enact stronger guardrails around the technology over concerns it could be used to overcome “knowledge barriers” that have historically prevented bad actors from creating biological weapons.“We believe it would be good for the world to have the option to slow or temporarily pause frontier AI development to enable societal structures and alignment research to keep up with the advance of the technology,” Favaro and Clark said.“But if a slowdown simply lets the least cautious actors catch up technologically, it could leave everyone less safe. Without a global coordination mechanism, companies and governments will have to make difficult decisions about safety while under competitive and geopolitical pressures.”AI agents are becoming increasingly popular, including among crypto users. Some crypto executives have speculated that AI agents settling transactions could drive adoption and transaction volumes. Circle CEO Jeremy Allaire predicted in January that billions of AI agents would operate on users’ behalf within five years.Crypto investment firm Keyrock reported last month that AI agents settling payments went from concept to reality in the past 12 months, with $73 million settled across 176 million transactions.Magazine: Korea’s first memecoin rug-pull case, China’s crypto rules review: Asia Express 

Crypto exchange Coinbase said it froze more than $3 million in cryptocurrency tied to a global operation targeting cyber-enabled crypto scam networks in Southeast Asia. The operation was part of Disruption Week led by the US Department of Justice’s Scam Center Strike Force, which brought together government entities and private industries to tackle crypto fraud targeting Americans. “This operation is proof that scammers can’t be stopped by any single company or agency acting alone,” said Coinbase. “It took social platforms, financial institutions, connectivity providers, and law enforcement working in lockstep to hit these networks at nearly every point in the fraud chain, online accounts, financial flows, and physical infrastructure all at once.”The operation also involved Meta, Microsoft and Starlink, which worked together to take down servers and other hosting infrastructure linked to scam networks and disrupt criminal activity across more than 1.4 million social media and email accounts, leading to several arrests by the Royal Thai Police Anti-Cyber Scam Center.Investment fraud and pig butchering are among the fastest-growing and most financially devastating forms of fraud targeting Americans, the DOJ said. The FBI reported earlier this month that Americans’ losses from crypto- and AI-related scams in 2025 exceeded $11 billion, with investment scams the most damaging.Source: US Attorney DC“We know crypto sometimes gets an unfair reputation when it comes to illicit finance. The reality is the opposite. Blockchain technology gives law enforcement something traditional financial systems often can’t: a transparent, immutable and permanent record of every transaction,” Coinbase added. Other members of the coalition included the FBI, the US Secret Service and law enforcement partners in the UK, Australia, Canada, New Zealand and Thailand.Related: OFAC sanctions Cambodian politician linked to pig butchering scam centers Meta said it shared “actionable insights” from all those involved, which helped “connect the dots between disparate pieces of information across platforms. The collective effort of these tech companies enabled the operation to target and disrupt criminals at nearly every point in the fraud chain.” Authorities around the world have been heavily targeting scam infrastructure this year. In April, the US Scam Center Strike Force and its law enforcement partners targeted scam centers and froze more than $701 million in crypto linked to investment scams.Meanwhile, a Dubai police-led international crackdown on scam rings resulted in the arrest of 276 individuals and the shutdown of at least nine crypto scam centers.In another police action involving Austrian and Albanian authorities, with support from Europol and Eurojust, 10 people were arrested in connection with three scam centers in Tirana, Albania.Magazine: Big Questions: Do we really only need 2–5 cryptocurrencies?  

​The European Banking Authority and the New York State Department of Financial Services (NYDFS) have signed a memorandum of understanding to police cross-border stablecoin activities. The EBA said on Tuesday that the deal is part of its duties under the Markets in Crypto-Assets (MiCA) Regulation and sets out principles and procedures for exchanging information and coordinating stablecoin supervisory activities, market trends, and risks between New York and the European Union.NYDFS said the deal would “enhance the supervision of entities engaged in stablecoin activities, identify market trends and risks, and promote the integrity of the stablecoin market.” Banks and major financial institutions in the US and Europe have tested using stablecoins for payments, spurred on by laws regulating the tokens in the US and EU. The global stablecoin market has grown to more than $319 billion as of Wednesday, according to DefiLlama.Source: European Banking Authority Some of the information the two watchdogs will share includes the issued stablecoin, total volume in circulation, the number of holders, results of external and internal audits and the regulatory standing of specific products and services.The MOU also provides a framework for the two regulators to assist each other and coordinate efforts during crises or emergencies. However, only supervised entities’ stablecoin-related activities will be monitored, not all activities a company might conduct.Related: ‘Stablecoins’ are an outdated term from crypto’s early years: A16zUS President Donald Trump signed stablecoin regulations into law in July, while the European Union’s Markets in Crypto-Assets framework came into effect toward the end of 2024. US dollar-denominated stablecoins currently make up the lion’s share of activity in the sector, with Tether’s USDT and Circle’s USDC the two largest by market capitalization. Jimmy Xue, co-founder of quantitative yield protocol Axis, told Cointelegraph in January that the global stablecoin market has largely plateaued after rapid expansion, entering a consolidation phase as new regulation, liquidity constraints, and higher real-world yields weigh on new issuance.Xue added that a cautious macroeconomic environment, combined with competitive Treasury yields, further reduced appetite for rapid stablecoin expansion.Magazine: Big Questions: Do we really only need 2–5 cryptocurrencies?  

Modern artificial intelligence-powered robots are impressive in their capabilities, but are still years away from replacing humans as they can’t yet adapt to changing conditions, researchers say. Last month, AI robotics company Figure showcased its humanoid robots performing basic tasks, such as cleaning a room, but a series of robots working for nine days straight sorting packages sparked conversation about how soon robots could replace jobs. Oliver Obst, an associate professor of robotics at the Australia based University of New South Wales, told Cointelegraph that repetitive jobs such as physical work in structured environments are currently most at risk of being replaced by robots, while administrative and document-processing tasks could be replaced by AI.There has been growing concern that AI and robots will replace people in jobs as technology advances. A report in May from workforce consulting firm Challenger, Gray and Christmas found that US companies have laid off an estimated 49,135 people in 2026 due to AI.A group of Figure’s robots worked for nine days straight sorting packages. Source: Figure However, Obst said that humanoid robots are unlikely to see a mass rollout soon because they don’t appear to be more efficient or less error-prone than current robotic manufacturing methods.“Even in relatively structured settings, they still face problems with reliability, speed, safety, cost, and recovery from unexpected situations,” he said. “The harder the environment is to control, the harder the robotics problem becomes. Most human jobs involve more variation and more judgment than the package-sorting demonstration.”“I would not say we are at the point of mass replacement by humanoid robots. We are much closer to the selective automation of some tasks. AI software is moving faster and is already affecting some forms of information work, but physical robots still have a much harder problem to solve.”In another video in May, a human worker managed to sort more packages compared to a team of Figure’s robots, which swapped out when needing a recharge. Figure CEO Brett Adock said it would be the last time “a human will ever win.”Source: Brett AdockPeople still better than bots in some areas Markus Levin, co-founder of decentralized data network XYO, said AI models and automation software can perform repetitive tasks with far greater consistency and endurance than humans; however, robots still require charging, maintenance and supervision.A report in September from the International Federation of Robotics found that global demand for factory robots has doubled over the last decade, with warehouses and logistics among the fastest-growing areas of adoption.“I believe broad human replacement is still likely years away,” Levin added, “Reliability, safety, regulation, infrastructure costs, and trust remain major barriers to full-scale deployment across society. The challenge is no longer simply making machines capable of acting but ensuring they can operate safely and reliably as they take on greater autonomy.”Dr Francisco Cruz Naranjo, a senior lecturer at the University of New South Wales with a PhD in robotics, said the efficiency of robots compared to people depends heavily on the activity and the environment.Related: ‘Developed ecosystem’ based on crypto has sprung up for AI agents: Report“Robots are much better at repetitive tasks without the need for constant pauses, as showcased in the Figure livestream. However, in highly dynamic environments, robots still struggle to quickly adapt to changing conditions,” he said.“Humans, in this case, are much better. This is precisely why robots at the moment are highly efficient in controlled environments, such as factories, but they have not yet succeeded widely in home settings.”Naranjo said repetitive jobs performed in a less static setting are at risk of being replaced by robots, but it will depend on how quickly research advances and how quickly society adapts in areas like making spaces robot-friendly, which is likely years away.Robots in society could be beneficial Naranjo and Obst said that a mass rollout of robots in the workforce could be of some benefit, such as improving work-life balance, increasing the workforce in areas with shortages, and addressing dangerous environments that are too risky for humans.“The social question is harder. If robots make dangerous work cheaper in human terms, that can be good. But it can also have unintended consequences. For example, keeping humans out of harm’s way in military operations may save lives, but it could also lower the perceived cost of conflict,” Obst said. “Hypothetically, if we became very successful at automating almost all work, then society would need to rethink economies that are currently built around individual wages and employment.”Magazine: Korea’s first memecoin rug-pull case, China’s crypto rules review: Asia Express

Stock and crypto trading platform Robinhood has entered the Canadian market after closing a $180 million stock acquisition of local crypto technology company WonderFi, gaining the company’s licenses and regulatory approvals in the country.Robinhood said on Monday that the acquisition marks the platform’s entry into Canada and will see crypto exchanges Bitbuy and Coinsquare come under the company’s banner.Bitbuy and Coinsquare are among Canada’s largest crypto exchanges. WonderFi disclosed in March that the platforms generated combined revenue of $49.8 million in 2025.Johann Kerbrat, the general manager of Robinhood Crypto and International, said in a statement that WonderFi has “extensive experience operating regulated crypto platforms that serve beginner and advanced crypto users alike, making it an ideal partner to accelerate Robinhood’s mission in Canada.”Source: Vlad TenevCrypto payments company Triple A estimated that roughly 4.1% of Canadians own crypto, while the analytics and consulting firm Grand View Research estimated that the Canadian crypto market generated around $263 million in revenue in 2025, driven by hardware. The firm also said Canada is the fastest-growing regional market in North America and projects total revenue will surpass $1 billion by 2033.Robinhood said WonderFi’s employees, including the leadership team, will stay on as part of the deal. It also expects to gain about 300,000 funded customers from WonderFi.Related: Robinhood announces $1.5B share buyback as stock struggles in 2026WonderFi and Robinhood first entered into an agreement last May at 36 Canadian cents per common share ($0.26). WonderFi’s stock has been drifting between 34 and 36 Canadian cents for the last month, according to Google Finance.Robinhood entered the US crypto trading market in February 2018. In April, the company was tapped to serve as brokerage and initial trustee for Trump Accounts, a new tax-advantaged investment account program for children. The company also launched an Ethereum layer-2 network on a testnet in February with plans for a mainnet launch later this year. Robinhood CEO Vlad Tenev said the network processed 4 million transactions in its first week of public testnet activity.Magazine: HYPE chases $100 target, ETH could dump below $1800: Market Moves Cointelegraph is committed to independent, transparent journalism. This news article is produced in accordance with Cointelegraph’s Editorial Policy and aims to provide accurate and timely information. Readers are encouraged to verify information independently.