Helen Toner joined the board of OpenAI in 2021.

There is significant political pressure to weaken the transparency requirements currently being developed for the EU AI Act.

Helen Toner prefers "conditional" AI development slowdowns, where progress is gated on achieving specific levels of system understanding or risk mitigation, over input-based slowdowns like compute cap...

Companies like Palantir and Scale AI have advertised large language model-based systems that mix sensible military uses with highly problematic ones, such as generating "non-escalatory" courses of act...

Even if AI ultimately helps cyber defenders more than attackers, there will likely be a dangerous transition period where attackers can leverage new models much faster than critical infrastructure pro...

Under Xi Jinping's leadership since 2012, China has become more illiberal, more aggressive in the South China Sea, and has made it clear it does not want to be part of the existing "rules-based intern...

The EU AI Act is developing a code of practice that includes transparency requirements for models trained with over 10^25 floating-point operations.

The US House of Representatives is currently less functional than it has been at any time since the post-Civil War era, which is a major barrier to passing AI legislation.

The rumor that a research breakthrough known as Q* contributed to the OpenAI board's decision to fire Sam Altman is "totally false."

Helen Toner believes the most effective way to implement AI whistleblower protections is to pair them with mandatory disclosure requirements for companies.

The US government's 10^26 compute threshold for reporting large AI training runs is on "wobbly footing" and its future is unclear.

AI policy should focus on using the "adaptation buffer"—the time between a capability's frontier creation and its widespread proliferation—to build societal resilience rather than attempting non-proli...

The US is currently on a trajectory for a "massive knee-jerk overreaction" in AI regulation following the first major AI-related incident.

A major source of current US-China hostility is the failed attempt in the 1980s-2000s to integrate China as a "responsible stakeholder" into the US-led international order.

The shift in rhetoric from AI CEOs to a more hawkish stance on China is likely a strategic move to secure funding, government contracts, and favorable regulation by appealing to the bipartisan consens...

The Reuters story claiming the OpenAI board's decision was based on a letter from employees about a research breakthrough was "totally false."

Deepseek matched the performance of US AI models that were 6 to 9 months old at a lower price point, illustrating how AI capabilities become cheaper over time.

An "AlphaGo for war" scenario is not a near-term possibility because real-world warfare is an intractably complex simulation problem with too many unknown variables and adversarial actions to be suita...

The US military has used the Aegis missile defense system, an autonomous weapon system developed in the 1960s-70s, since the 1980s to automatically identify and engage incoming missile threats.

A potential model for AI regulation involves requiring companies to create and publish a safety and security plan, which would then create a clear basis for whistleblowing if the company fails to adhe...