Trump said on September 14 that additional safeguards were unnecessary and dismissed warnings from technology executives about uncontrolled development. His position places the White House at odds with industry figures who have urged stronger testing, independent audits and tighter controls on advanced models capable of autonomous cyber activity.
The dispute has immediate relevance for digital assets because AI tools are increasingly being used on both sides of the security contest. Developers deploy them to review smart-contract code, identify suspicious transactions and automate monitoring, while attackers can use the same capabilities to improve phishing, impersonation, reconnaissance and exploitation at greater speed.
A June executive order from Trump’s administration prioritised enforcement of federal laws against people who use AI to gain unauthorised access to computer systems or commit related crimes. The order, however, emphasised innovation and rejected what it described as burdensome regulation, leaving much of the responsibility for risk controls with companies and existing agencies.
That approach is now being tested by a sharp rise in cryptocurrency attacks. CertiK recorded more than $1.31 billion in Web3 security losses across 344 incidents during the first half of 2026, with wallet compromises accounting for more than $444 million. TRM Labs separately counted 207 hacks in the same six-month period, the highest number it had recorded for any half-year, while putting total losses at $972 million.
Differences between industry datasets reflect varying definitions and methodologies, but they point in the same direction: more attacks, operational complexity and a growing focus on infrastructure rather than only flaws in smart contracts.
Fireblocks said several attacks during the second quarter demonstrated how social engineering and weak operational controls could be combined with AI-enhanced techniques. Its assessment of an April breach involving Zerion said attackers used AI-generated personas and interaction methods to build trust before obtaining credentials and keys from internal systems. User funds were protected by the self-custodial structure, but the case illustrated how generative AI can strengthen intrusion methods without requiring a novel technical exploit.
Security researchers are also watching the development of autonomous agents that can chain together multiple steps, including reconnaissance, credential theft and system probing. Anthropic said in a September threat assessment that misuse of advanced models had moved beyond isolated prompts towards more agentic workflows capable of carrying out portions of cyber operations with limited human supervision.
That shift matters for DeFi, where open-source code, automated contracts and rapidly moving liquidity can create narrow response windows. Once a vulnerability is identified, an attacker can often execute transactions globally within minutes, while defensive teams may need governance approval, exchange co-operation or validator action before funds can be frozen or redirected.
Trump’s stance does not remove existing cybercrime, fraud or market-manipulation laws, and it does not prevent regulators or private companies from imposing their own security requirements. It does, however, reduce the likelihood of federal rules specifically mandating safety testing or deployment limits for frontier AI systems unless Congress acts independently.
Senators from both parties are discussing legislation that could require major AI developers to demonstrate protections against catastrophic misuse and permit government testing of advanced systems. The proposal remains under debate and faces uncertain prospects without White House support.
AI company leaders have called for stronger safeguards despite differences over how they should be structured. Anthropic chief executive Dario Amodei has advocated independent evaluations of powerful systems. OpenAI chief executive Sam Altman has backed oversight for frontier models, while Elon Musk has repeatedly warned that advanced AI could create risks if development outpaces controls.
Follow Arabian Post
Select Arabian Post as your preferred source on Google and MSN News for trusted business news and Arab politics and updates.
