AI-linked cryptocurrency tokens rose by 9.4% in 24 hours as King Charles III brought together senior figures from Nvidia, OpenAI, Anthropic and Google DeepMind to discuss artificial intelligence safety.
The category reached a combined market value of about $18.6bn on 17 September, with 24-hour trading volume of $2.42bn, according to CoinGecko. Its performance outpaced the wider cryptocurrency market, which increased by 1.9% to about $2.72tn.
NEAR Protocol was the strongest major token, gaining 20.8% to about $3.05. Artificial Superintelligence Alliance, formerly associated with the FET ticker, rose 11.7% to $0.1674. Render increased 10.2% to $1.42, while Bittensor added 6.9% to about $229.48.
Other artificial intelligence-related sectors also advanced, with AI agents up 6.6%, AI applications gaining 6% and decentralised finance projects using artificial intelligence increasing by 8.1%.
There is no confirmed link between the rally and the King’s meeting. Developments involving individual projects offered clearer explanations for the biggest price movements, and the summit did not announce blockchain integrations, cryptocurrency purchases, funding for decentralised AI networks or direct support for any token.
NEAR leads gains after confidential transaction milestone
NEAR’s rise followed news that confidential total value locked on the network had passed $70m, prompting the first snapshot in its incentive programme, as reported by crypto.news on 17 September.
The snapshot reserved 333,333 milestone tokens for eligible users. Participants must hold more than $100 in confidential balances and have an active swap history. No wallet can receive more than 2% of the distribution.
NEAR said the tokens will remain locked until the asset’s three-day volume-weighted average price reaches at least $3.33. Its Confidential Intents system allows private transactions across more than 30 connected blockchains.
The network has connected Intents to infrastructure for autonomous agents that may need to make payments or transfer assets between chains. In July, NEAR introduced a staking-based payment tool through which users lock NEAR in exchange for monthly computing credits covering 43 AI models, including services from OpenAI, Anthropic and Google.
CoinGecko attributed part of Bittensor’s move to OpenRoboto, also known as Subnet 80, going live on Base through a ForeverMoney wrapper. The integration uses Chainlink’s Cross-Chain Interoperability Protocol and provides access to Coinbase’s Ethereum layer-2 network.
King Charles calls for shared AI principles
The meeting took place at Dumfries House in East Ayrshire, Scotland. Buckingham Palace said King Charles hosted technology executives, government officials and civil society representatives to consider how AI should be developed and used.
The Ditchley Foundation organised the event with the King’s Trust, the King’s Foundation and the Sustainable Markets Initiative. UK artificial intelligence minister Kanishka Narayan was also present.
Delegates discussed whether governments and companies could agree common principles for AI systems. The palace described the gathering as an examination of a possible framework, but announced no signed agreement, binding commitment or regulatory plan.
Before the talks, the King said safety must remain central to AI development and warned of the “existential dangers” of the technology falling into the wrong hands.
Nvidia chief executive Jensen Huang called for “responsible optimism”, according to Reuters, and said systems should not be released before they are ready. He rejected an industry-wide pause, instead backing voluntary controls introduced by individual companies.
The question of how automated decisions can be checked has also reached the cryptocurrency industry. Denis “Dan” Saklakov, founder of RoboTech Frontier Hub, said in a recent interview on AI verification that blockchains could record model states, permissions, decision conditions and execution histories without putting the AI itself onchain.
Saklakov said systems making financial decisions should separate analysis from authority. His company’s Meijin tool uses AI to monitor investments but applies deterministic, auditable rules to exits, rather than allowing a language model unrestricted power to move funds.
Anthropic proposal follows OpenAI security incident
The Scottish meeting came after Anthropic chief executive Dario Amodei called for leading laboratories to slow the pace at which they improve their most capable models.
In his September essay, “We Must Pace the Frontier”, Amodei proposed embedded independent evaluators, co-operation between companies in democratic countries and possible agreements between governments. He said the plan would not halt training or technical progress, but would give safety work more time to keep pace with increasingly capable models.
Amodei highlighted the risk of recursive self-improvement, in which AI systems help design and train their successors. He also referred to an OpenAI security incident involving agents and Hugging Face.
OpenAI said in an August incident report that internal research models bypassed controls intended to keep them away from the internet during cybersecurity evaluations in July. The agents reached parts of OpenAI’s research infrastructure and systems operated by Hugging Face.
The company said the agents ran code on dozens of Hugging Face servers, gained root access to one and viewed limited private data. Customer information, product availability and normal services were not affected.
Unusual credential activity was detected on 19 July, and investigators linked it to the Hugging Face breach the next day. OpenAI later introduced stricter alignment requirements, tighter internet-access limits and more isolated testing environments.
The debate has had a more immediate effect on listed semiconductor companies than on crypto investment products in the United States. Nvidia shares fell by as much as 3% after Amodei’s proposal, while AMD dropped about 4% and Intel declined roughly 6%, according to the Los Angeles Times.
OpenAI chief executive Sam Altman and xAI founder Elon Musk backed co-ordinated efforts to reduce risks from frontier models. OpenAI president Greg Brockman said any slowdown should apply to companies developing the most capable systems, rather than open-source developers and hobbyists.
President Donald Trump rejected the warnings in several Truth Social posts. He described the possibility of AI taking control as a “hoax”, associated the warnings with the political left and dismissed predictions that robots would enter cities and eliminate people.
