Technology & Science
California AG Orders xAI to Halt Grok’s Sexual Deepfake Output
On 17 Jan 2026, Attorney General Rob Bonta served xAI a cease-and-desist letter giving the firm five days to stop Grok from generating and distributing non-consensual sexual images, invoking California’s brand-new deepfake law AB 621.
Focusing Facts
- Research cited by Bonta found that >10,000 of roughly 20,000 Grok images produced between 25 Dec 2025 and 1 Jan 2026 showed scantily-clad or apparently under-age subjects.
- AB 621, in force since 1 Jan 2026, lets prosecutors seek civil awards up to $250,000 and fines of $25,000 per violation against platforms that ‘recklessly aid’ deepfake distribution.
- Malaysia, Indonesia and the Philippines blocked Grok outright in early Jan 2026, and Japan opened its own probe the same day California acted.
Context
California’s action echoes earlier tech flashpoints where states moved faster than Washington—think 2004’s auto-emissions rules that forced national standards or 2018’s GDPR in Europe shaping global privacy norms. A tighter historical rhyme is the 1890 public outcry over Kodak’s handheld camera, which birthed modern privacy torts; again a novel technology made it suddenly easy to create unwanted, intimate images, and state courts stepped in before federal law. Today’s cease-and-desist signals that the laissez-faire phase of generative-AI is closing as regional regulators weaponise existing child-protection and obscenity statutes instead of waiting for bespoke federal rules. Over a 100-year arc, the episode may mark the inflection where AI companies lose the Section 230-style shield and must bake safety constraints into model design, much as radio learned spectrum discipline after the 1934 Communications Act. Whether this becomes a lasting template or another skirmish overturned on First-Amendment grounds will shape the balance between innovation speed and the centuries-old legal duty to prevent the exploitation of the vulnerable.
Perspectives
Regional California media
Regional California media — Portrays Attorney General Rob Bonta’s cease-and-desist order as an essential legal step to stop Grok from spreading illegal, non-consensual sexual deepfakes of women and children. Close proximity to state politics and reliance on official statements may lead these outlets to emphasize regulatory success and public danger while giving scant attention to free-speech or technological complexities.
Tech industry outlets
Tech industry outlets — Frames the crackdown as a high-stakes confrontation between Musk’s ‘truth-seeking’ AI ambitions and regulators worried about harm, underscoring the wider tension between innovation and government control. Writing for a tech-centric audience, coverage often echoes Silicon Valley narratives about over-regulation and free expression, potentially downplaying the lived impact on victims of deepfakes.
Crypto and blockchain-focused media
Crypto and blockchain-focused media — Uses California’s action against xAI alongside Oklahoma’s new bills to depict a growing wave of state-level AI laws challenging federal limits and targeting deepfakes. Libertarian, market-oriented leanings drive the focus toward jurisdictional power struggles and business implications, sometimes glossing over the specific ethical and safety issues underlying the controversy.