Joe Miller in Washington and Cristina Criddle in San Francisco
 
 
Anthropic will offer its Claude chatbot to US lawmakers for a nominal $1 fee, as the artificial intelligence start-up battles for influence in Donald Trump’s Washington.
In a deal agreed with the government on Tuesday, the group followed arch-rival OpenAI in offering its enterprise tools to the executive branch for $1 per federal agency, while also agreeing to an equivalent discount for senators, members of Congress and top judges. 
“We need to get widespread adoption [of AI tools] in the federal government,” said Josh Gruenbaum, the commissioner of the Federal Acquisition Service, which co-ordinates procurement for US agencies. “The price is going to help uptake from agencies happen that much quicker.”
The three branches of the US government generally have their own procurement processes and no agency or organisation would be obliged to use Claude under the agreement. The Claude for government product is permitted to be used by federal workers for sensitive, unclassified work.
Google was also in talks with the government to offer its Gemini chatbot to federal employees on similarly cheap terms, a person familiar with the negotiations said. Google did not immediately respond to a request for comment.
The Anthropic and OpenAI deals, which will initially last one year, highlight how industry leaders are offering ever-greater incentives as they compete for widespread adoption by the government and attempt to deepen ties to lawmakers and regulators.
“America’s AI leadership requires that our government institutions have access to the most capable, secure AI tools available,” said Anthropic chief executive Dario Amodei. “By offering expanded Claude access across all three branches of government, we’re helping the federal workforce leverage frontier AI capabilities.”
The government approved Anthropic’s Claude, Google’s Gemini and OpenAI’s ChatGPT as suppliers earlier this month. It is also looking to ink deals with other “frontier” AI models, such as Meta’s Llama and xAI’s Grok, according to a US official, as well as with smaller, bespoke artificial intelligence platforms.
While tech companies do not generate significant revenue from such deals, they benefit from learning which applications are most popular at different agencies. This could lead to products being sold to other enterprise clients or fresh deals after current arrangements expire.
The approval of Claude, Gemini and ChatGPT comes after the White House last month vowed to stop companies whose systems exhibit “partisan bias or ideological agendas” from doing business with the US government. Trump’s political allies have long accused models such as ChatGPT and Gemini of promoting liberal ideology.
A US official cautioned that the approval of the platforms did not preclude them from being reviewed for alleged bias in the coming months.
Gruenbaum also stressed that deals with OpenAI and Anthropic did not signal the government’s preference for a particular model.
“We’re not picking a winner here,” he said. “There’s almost a market efficiency when you have that competitive tension and dynamic to make these tools accurate and truthful.”
OpenAI and Anthropic are burning through cash at historic rates as they spend tens of billions of dollars on the computing power needed to train and run AI models and fund lucrative pay packages amid a talent war. They are supported by soaring valuations and hordes of willing investors despite reporting huge annual losses.
OpenAI is in talks for an employee share sale at a $500bn valuation, while Anthropic is raising $5bn at a $170bn valuation. ChatGPT claims 700mn monthly active users, with Gemini trailing with 450mn.