Technology industry groups are pushing back against a Trump administration proposal to ban “woke” artificial intelligence in government contracting, calling the draft requirements technically infeasible and warning they could deter companies from doing business with Washington.
The General Services Administration, the government’s main procurement agency, has formally proposed requiring contractors to adhere to a broad list of “unbiased AI principles” in the systems they sell to the government, including “historical accuracy” and neutrality, NOTUS reported.
Under the draft guidance, the government would be able to test AI systems for bias without disclosing how it does so, and contractors would have to pay decommissioning costs if their contract is terminated as a result of such testing.
The Business Software Alliance, a trade group whose members include many government contractors, called the proposal “infeasible” in a comment submitted to the GSA. “Compliance with the proposed clauses could result in reduced performance and lower accuracy, which could itself result in biased outcomes, subjecting the government to legal liability,” the group wrote, arguing that large language models are “inherently probabilistic.”
The Information Technology Industry Council, a global tech trade association whose members include Google and OpenAI, is asking the administration to drop the “unbiased AI” requirement altogether. “Without clear definitions and transparent compliance benchmarks, contractors will be held to an impossible performance standard,” Megan Petersen, the group’s counsel and senior vice president of public sector policy, told NOTUS, warning the requirements could “discourage innovative commercial companies from doing business with the U.S. government altogether.”
Lawyers and procurement experts question how such a standard can be enforced. “How, in a world in which your political viewpoint colors what you believe to be the truth, is that enforced in the contract?” asked Jessica Tillipman, associate dean for government procurement law studies at George Washington University, describing the framework as mushy.
The proposal is the GSA’s second attempt at drafting contracting language for AI; a previous version required contractors to make “commercial efforts” to ensure compliance. The push follows the administration’s high-profile dispute with Anthropic, which was labeled a “supply-chain risk” and barred from federal use, a decision courts have questioned; a federal judge said this week that the administration still lacks evidence for the label, TechCrunch reported.
Supporters of the effort argue the government is setting ground rules for AI adoption. “The government is just laying out the ground rules ahead of time,” Joel Thayer, a senior fellow for AI and emerging technology policy at the America First Policy Institute, told NOTUS. With the administration seeking to ramp up AI use across federal agencies, the clash over “woke AI” language highlights how difficult it is to translate political demands into procurement rules.
Sources: NOTUS, NPR, TechCrunch
Author: Technology Desk
discussion