President Joe Biden signed a broad executive order on Monday clamping down on false content produced by artificial intelligence (AI), a move experts say could open the door to more online censorship.
The Biden administration recently secured “voluntary commitments” from leading technology companies in July to address the risks posed by AI and it also released a blueprint for an AI bill of rights that is largely about “discrimination,” “equity” and “bias.” Monday’s executive order appears to focus in part on safeguarding Americans by labeling and identifying deepfakes, which are fraudulent content created by AI that appears genuine, according to a fact sheet outlining the order.
While there is an absence of explicit references to “misinformation” or “disinformation” in the fact sheet, the vague language leaves open the possibility for content censorship, experts told the DCNF.
“The executive order’s justifications leave open the possibility for all sorts of things — especially censorship,” Digital Progress Institute President Joel Thayer told the DCNF. “If we are simply placing labels on things, then what constitutes an AI-label? What are regulatory actions they can take if not appropriately labeled? Do they force companies to take down content? How much influence will the White House have over those decisions?”
An example of clearly identifying content created by AI is “watermarking” it, according to the fact sheet.
Government regulation of AI is important but it can go too far, Director of Policy for American Principles Project Jon Schweppe told the DCNF.
“There’s a role for direct government oversight over AI, especially when it comes to scientific research and homeland security,” Schweppe explained. “But ultimately we don’t need government bureaucrats micromanaging all facets of the issue.”
Experts also raised concerns about the executive order’s guidance on preventing “algorithmic discrimination,” which is when AI systems do not produce equitable results in areas such as housing, health care and criminal justice.
“Certainly we shouldn’t want a Bureau of Artificial Intelligence running around conducting investigations into whether a company’s AI algorithm is adequately ‘woke,'” Schweppe asserted.
The order will “protect Americans from AI-enabled fraud and deception,” according to the fact sheet. The Department of Commerce will establish the recommendations for identifying “AI-generated content and authenticating official content.”
The words “authenticating official content” raised a red flag for R Street Institute Senior Fellow Adam Thierer.
“Today’s Executive Order invites a lot of potential meddling by federal bureaucracies in AI markets and that includes the market for speech,” he told the DCNF. “Government efforts around ‘authenticating official content’ raise thorny questions about what constitutes disinformation or misinformation from a regulatory perspective.”
“We’ve already witnessed some policy fights about the Department of Homeland Security’s (DHS) controversial Disinformation Governance Board and we can expect those squabbles to intensify as more bureaucrats seek to make determinations about what constitutes acceptable algorithmic speech,” Thierer added.
DHS Secretary Alejandro Mayorkas told lawmakers in 2022 that the agency had formed the Disinformation Governance Board to combat misinformation and disinformation targeting minority communities. The board received broad scrutiny from Republicans and the DHS paused it.
Biden’s Assistant Commerce Secretary for Communications and Information Alan Davidson asserted in April that AI regulation could include government evaluations of whether AI is spreading “misinformation, disinformation, or other misleading content.”
Post a Comment