Deepfakes and the Psychology of Courtroom Belief - Psychology Today

In a case that sent shockwaves through the legal world, a California judge narrowly identified an AI-generated deepfake submitted as testimony by plaintiffs. While the judge successfully spotted technical flaws such as looped video and repeated eye blinks, experts warned that the technology is rapidly evolving beyond such detectable errors, leaving the legal system scrambling to adapt.

What happened

The incident involved the submission of fabricated video evidence, which historically has been treated by courts as a reliable, incontrovertible "smoking gun." The judge's ability to detect the forgery was attributed more to luck than to systematic safeguards. This event highlights a broader crisis: the "veridicality heuristic," where human beings intuitively equate visual perception with truth. Because humans process visual information faster and with more emotional weight than text, fabricated videos depicting false confessions or actions exert a powerful, misleading psychological pull on jurors.

The threat is compounded by a phenomenon known as the "deepfake defense," where the existence of AI synthesis allows bad actors to dismiss legitimate recordings as fabrications. This creates a psychological double bind in the courtroom, potentially leading to epistemological paralysis where jurors struggle to trust any evidence. The real-world consequences are already manifesting; in a separate incident noted in February 2026, an unnamed woman spent two days in jail after her ex-boyfriend allegedly fabricated AI-generated text messages to trigger her arrest for violating a protective order. Although charges were eventually dropped, the ordeal required eight months of legal proceedings. As generative AI continues to improve through feedback loops between content-generating and evaluating algorithms, the legal community is now facing an arms race that human perception is poorly equipped to win.