Nvidia CEO Jensen Huang has warned that China already has the infrastructure and computing power needed to train advanced AI models comparable to Anthropic’s Claude Mythos, raising concerns about potential cybersecurity risks.

Speaking on the Dwarkesh Patel podcast, Huang said the level of compute used to train Mythos is not particularly rare and is already widely available in China.

China’s AI Infrastructure Is “Abundant”

Huang emphasized that the type of hardware and capacity required to build a model like Mythos is not out of reach for China.

“The amount of capacity and the type of compute it was trained on is abundantly available in China,” he said, adding that the country already has access to the necessary chips and infrastructure.

He pointed out that China has significant unused data center capacity, describing it as having “enormous” compute resources, including fully powered but underutilized facilities.

According to Huang, China’s broader advantages include producing around 60% of the world’s mainstream chips, having a large share of global AI researchers, and access to substantial energy resources.

Rising Concerns Over AI and Cybersecurity

The warning comes amid growing concerns about the capabilities of Anthropic’s Claude Mythos model.

The AI system has demonstrated the ability to identify thousands of software vulnerabilities across major operating systems and browsers. Reports suggest that a large portion of these vulnerabilities remain unpatched, increasing the risk of exploitation.

Security researchers have also found that the model can autonomously execute complex, multi-step cyberattacks, tasks that would typically take human experts days to complete.

If a similar model were developed and misused, it could pose serious risks to global cybersecurity, particularly for institutions relying on outdated systems.

Call for Cooperation Over Confrontation

Despite the concerns, Huang cautioned against treating China purely as an adversary.

While acknowledging geopolitical tensions, he argued that collaboration and dialogue around AI development may be a more effective approach to managing risks.

“We want the United States to win,” Huang said, “but having research dialogue is probably the safest path forward.”

US Officials Highlight AI Competition

Meanwhile, US Treasury Secretary Scott Bessent recently described Claude Mythos as a major leap forward in AI capabilities, suggesting it strengthens the US position in the global AI race.

However, the rapid pace of development on both sides underscores the competitive and high-stakes nature of the sector.

Growing Evidence of AI Misuse

Concerns about misuse are not purely theoretical. Anthropic previously reported that a China-linked group attempted to exploit its AI coding tools to target dozens of global organizations, succeeding in some cases.

As AI systems become more powerful and accessible, experts warn that the line between innovation and risk is becoming increasingly thin.

The US Commodity Futures Trading Commission (CFTC) is reportedly investigating suspicious activity in oil futures markets tied to key announcements made by the Trump administration during the Iran conflict.

According to a Bloomberg report, the probe focuses on trading activity across major platforms, including CME Group’s NYMEX and the Intercontinental Exchange.

Unusual Trading Ahead of Key Announcements

Regulators are examining at least two instances where oil futures trading volumes spiked shortly before major policy updates.

The first occurred on March 23, when billions of dollars in futures contracts were traded roughly 15 minutes before President Donald Trump delayed planned strikes on Iranian energy infrastructure.

The second took place on April 7, just before Trump announced a two-week ceasefire with Iran.

In both cases, the surge in trading activity was followed by falling oil prices and rising equity markets, raising concerns about potential insider trading.

CFTC Seeks Trader Identity Data

As part of the investigation, the CFTC has requested “Tag 50” data from exchanges. This information helps identify the individuals or entities behind specific trades and is commonly used for compliance and auditing purposes.

The regulator is aiming to determine whether any traders had access to non-public information ahead of the announcements.

Broader Crackdown on Insider Trading

The investigation comes amid increased scrutiny of insider trading, particularly in both traditional futures markets and newer prediction markets.

Brian Young, a former CFTC enforcement director now at Jones Day, noted that regulators are highly motivated to pursue such cases, given the real-world impact of oil price movements.

“Prices at the pump closely correlate with oil futures,” he said, emphasizing that these markets directly affect consumers.

Prediction Markets Also Under Watch

CFTC enforcement director David Miller recently warned that insider trading rules also apply to prediction markets, pushing back against claims that such platforms operate outside traditional regulations.

“There’s a myth that insider trading doesn’t apply in prediction markets. That is wrong,” he said.

In response to mounting regulatory pressure, platforms like Kalshi and Polymarket have introduced new measures aimed at preventing insider trading.

Meanwhile, US lawmakers have introduced the Public Integrity in Financial Prediction Markets Act of 2026, which seeks to limit insider trading by government officials.

Increased Oversight Across Markets

The ongoing probe signals a broader effort by regulators to tighten oversight across financial markets, especially during periods of geopolitical uncertainty.

As investigations continue, the outcome could have implications not only for oil trading but also for how regulators approach transparency and fairness across both traditional and emerging markets.

BitMEX Research has put forward a new proposal aimed at addressing potential quantum computing threats to Bitcoin, offering an alternative to controversial plans that suggest freezing vulnerable coins.

Instead of immediately restricting access to older, quantum-exposed Bitcoin, the proposal introduces a more cautious “wait-and-see” approach centered around a so-called canary fund.

A Conditional Approach to Quantum Risk

The proposal suggests implementing a soft fork that would only trigger a freeze on vulnerable Bitcoin if a real quantum threat is proven to exist.

Rather than acting preemptively, the system would monitor for evidence that a quantum computer capable of breaking Bitcoin’s cryptography is actually in operation.

This stands in contrast to the recently proposed BIP-361, which recommended freezing dormant Bitcoin wallets that could be at risk in the future.

How the Canary Fund Works

At the core of BitMEX’s idea is a “canary address,” created using a cryptographic method known as a Nothing-Up-My-Sleeve Number. This ensures the address is valid but has no known private key.

Users can voluntarily send Bitcoin to this address, effectively creating a bounty. If a quantum-capable attacker is able to access and spend funds from this wallet, it would serve as clear proof that Bitcoin’s current cryptography has been compromised.

Only then would the proposed system activate a broader freeze on vulnerable coins.

This mechanism is designed to act as an early warning system, allowing the network to respond only when a real threat emerges.

Avoiding Immediate Restrictions

Under the “canary watch” model, existing Bitcoin holders would still be able to access and move their funds without restriction, as long as the canary address remains untouched.

Participants contributing to the fund could also use multisignature setups and retain the ability to withdraw their funds at any time.

The proposal also includes a safety window, allowing certain transactions involving older coins to continue even after a defined period, though with additional safeguards.

Response to Community Concerns

The idea comes after strong criticism of BIP-361, with some community members labeling the proposal as overly restrictive and potentially confiscatory.

BitMEX acknowledged that while its approach introduces added complexity, it may strike a better balance by minimizing disruption while still preparing for a worst-case scenario.

A Debate Still in Early Stages

Even the author of BIP-361, Jameson Lopp, has clarified that the proposal is more of a conceptual framework than a finalized plan.

He described it as a “rough idea for a contingency plan,” noting that while the concept may be unpopular, it was created to address a potentially serious future risk if quantum computing advances significantly.

For now, both proposals highlight an ongoing debate within the Bitcoin community about how to prepare for emerging technological threats without undermining core principles like decentralization and user control.