OpenAI CEO urges U.S. to prepare for AI ‘superintelligence’ risks and gains
The crypto industry faces growing cybersecurity risks as AI tools lower the cost and skill needed to exploit software flaws, with over $1.4 billion in assets stolen last year.
OpenAI CEO urges U.S. to prepare for AI ‘superintelligence’ risks and gains
The crypto industry faces growing cybersecurity risks as AI tools lower the cost and skill needed to exploit software flaws, with over $1.4 billion in assets stolen last year.
ChatGPT app open on iphone (Solen Feyissa/Unsplash)
What to know:
OpenAI CEO Sam Altman warned that advanced AI is moving into daily economic use, with next-generation models expected to help scientists make discoveries and enable individuals to do the work of entire groups.
The crypto industry faces growing cybersecurity risks as AI tools lower the cost and skill needed to exploit software flaws, with over $1.4 billion in assets stolen last year.
Altman highlighted potential threats including powerful cyberattacks and misuse of AI for biological research, urging urgent coordination across government and tech firms.
That shift is already visible in cybersecurity, where some industry leaders say artificial intelligence is tilting the balance toward attackers.
Charles Guillemet, chief technology officer at hardware wallet maker Ledger, for example, told CoinDesk that AI tools are lowering the cost and skill needed to find and exploit software flaws. Tasks that once took months, such as reverse-engineering code or linking multiple vulnerabilities, can now be completed in seconds with the right prompts.
The crypto industry saw more than $1.4 billion in assets stolen or lost in attacks last year. That figure could keep growing, Guillemet suggested. Moreover, developers are increasingly relying on AI-generated code, which may potentially introduce new flaws at scale.
The response, he said, will require stronger defenses such as mathematically verified code, hardware devices that keep private keys offline and a broader recognition that systems can fail.
AI in cyber, biosecurity
While Altman noted that AI could speed up drug discovery or materials science, he also flagged that it could also enable more powerful cyberattacks and lower the barrier to harmful biological research. Such threats may emerge within a year, which makes coordination across government, tech firms and security groups urgent.
“We’re not that far away from a world where there are incredibly capable open-source models that are very good at biology,” he said. “The need for society to be resilient to terrorist groups using these models to try to create novel pathogens is no longer a theoretical thing.”
Another example he suggested was a “world-shaking cyberattack” that could occur as early as this year. Avoiding that, he said, would require a “tremendous amount of work.”
He framed OpenAI’s policy ideas as a starting point, aiming to push debate on how to manage systems that learn fast and act across many fields. Using AI to help defend against these potential attacks, he said, is important.
On the potential nationalization of OpenAI, Altman said the case against it relies on the need for the U.S. to achieve “superintelligence” before its rivals do.
“The biggest case against nationalization would be that we need the U.S. to succeed at building superintelligence in a way that is aligned with the democratic values of the United States before somebody else does,” he said. “That probably wouldn’t work as a government project, I think that’s a sad thing.”
Still, Altman said he believes companies involved in AI must work closely with the U.S. government.
Given his role at OpenAI, Altman also has a financial stake in how the sector evolves. That position may shape how he frames both the urgency of regulation and the role of private companies like OpenAI in managing emerging risks, which could influence the firm’s competitive standing.
AI as a utility
Energy is one area where he sees quick progress because greater processing power capacity could keep costs down as AI demand grows.
Altman also pointed to early signs of labor shifts. A programmer in 2026, he said, already works differently to one a year earlier.
AI will become a sort of utility, like electricity, embedded across devices while the cost of basic intelligence falls and top systems remain expensive.
“You will have this personal super assistant running in the cloud,” Altman said. “If you use it a lot or use it at high levels of intelligence you’ll have a higher bill one month and if you use it less, you’ll have a lower bill.”
It’s “incredibly important that people building AI are high integrity, trustworthy people.”
Most crypto privacy models weaken as blockchain data grows. Encryption-based models like Zcash strengthen. CoinDesk Research maps the five privacy approaches and examines the widening gap.
Why it matters:
As blockchain adoption scales, the metadata available to machine learning models scales with it. Obfuscation-based privacy approaches are structurally degrading as a result. This report provides a comprehensive comparison of all five major crypto privacy architectures and a framework for evaluating which models remain durable as AI capabilities improve.
The win landed the same week listed miners Riot, MARA, and Genius Group disclosed selling more than 19,000 BTC from their treasuries.
What to know:
A solo bitcoin miner with roughly 230 terahashes per second of computing power unexpectedly validated block 943,411, earning 3.139 BTC worth about $210,000.
The miner, connected to the anonymous solo.ckpool.org pool and representing just 0.00002% of the network’s estimated 1-zetahash hashrate, faced roughly 1-in-28,000 odds of finding a block on…
