NeuralStackAI Law Radar’s late-July update cycle shows global AI regulation maturing in the details, with lawmakers and officials fine-tuning deepfake offences, algorithmic pricing bans, automated decision-making rules and grace periods rather than unveiling sweeping new frameworks. The service, which tracks every material change to its AI law register with dated, sourced entries, logged 35 recent updates ranging from new obligations to confidence-level upgrades. The result is a snapshot of a regulatory environment that is increasingly concrete about what AI can and cannot do in specific domains.
In the United Kingdom, the AI (Regulation) Bill in the House of Lords has quietly slipped further into the background as the government doubles down on a sandbox-first approach. AI Law Radar upgraded its confidence that the bill is unlikely to advance in its current form, citing UK Parliament’s Bills API showing no progress past first reading and the Department for Science, Innovation and Technology’s Blueprint for AI Regulation favouring an AI Growth Lab and sectoral sandboxes over a cross-cutting statute. The facts and dates for the bill remain unchanged, but its prospect of becoming the backbone of UK AI law looks weaker, reinforcing the view that sector regulators and non-legislative initiatives will carry more weight in the near term.
UK lawmakers are moving faster on the edges of AI harm, particularly around deepfakes and automated decisions. AI Law Radar added the Data (Use and Access) Act 2025 reform to automated decision-making, which inserts new sections into the UK GDPR and, from early February 2026, replaces the old default prohibition with a set of duties around notification, representation, human review and contest rights for people subject to automated decisions. It also added section 138 of the same Act, which came into force a day later and criminalises creating or even requesting a non-consensual intimate deepfake, whether or not it is ever shared. That offence is distinct from the Crime and Policing Act 2026’s focus on tool suppliers and, after a scope correction, is now tracked as binding on any person in the UK, not only AI deployers.
U.S. states are chipping away at algorithmic collusion in housing markets and reclassifying AI bans under clearer themes. Illinois SB 343, which amends the state’s Antitrust Act to prohibit algorithmic coordination of rental prices, has cleared the General Assembly and sits on Governor Pritzker’s desk with an action deadline in late August 2026, so AI Law Radar treats it as a proposed measure with no effective date yet. A related cleanup reclassified three state-level measures — New Jersey’s FAIR Act, Maryland’s Protection From Predatory Pricing Act and Illinois’s SB 343 — under a “Prohibited AI practices” theme after the service found they were previously tagged with undefined topics that made them hard to discover. All three share a core idea: banning certain uses of algorithms outright rather than merely imposing transparency or risk management duties.
The updates also highlight how technical accuracy and timing matter as much as substance in AI law. AI Law Radar corrected its citation for New York’s RAISE Act, a frontier AI safety law signed by Governor Hochul in December 2025, to the proper bill numbers while keeping the tracked effective date, and similarly fixed the location of Australia’s new automated decision-making transparency duty within the Privacy Act’s Australian Privacy Principles without altering the December 2026 commencement. In Rhode Island, law-firm trackers helped resolve uncertainty around the Healthcare AI Documentation Act’s timing, confirming it was signed in June 2026 and effective upon passage, which allowed the service to drop a placeholder note and raise its confidence level.
Further afield, regulators are pairing new duties with measured grace periods as high-impact AI rules come online. South Korea’s AI Basic Act, effective from January 2026, now has a confirmed at-least-one-year grace period on fines for high-impact AI systems, based on a government policy briefing quoting the Ministry of Science and ICT. That clarification did not change any tracked dates but did push the confidence rating to high, signaling that companies have some breathing room to adapt. Georgia’s Conversational AI Safety Act similarly had its effective date locked in via an official Senate press release, while Tennessee’s SB 1700 was substantively downgraded: amendments stripped out chatbot safety requirements before passage, leaving only a mandate for a state commission to study potential AI regulation with no compliance obligations and no report deadline.
China also features in the update stream, with AI Law Radar adding the CAC, NDRC and MIIT’s AI Agents Implementation Opinions issued in May 2026 to its register. While the changelog does not spell out the granular obligations in those Opinions, their inclusion signals that Chinese regulators are starting to codify expectations for AI agents, an area that blends large models with autonomy and anthropomorphic interfaces. This sits alongside other, previously tracked Chinese measures and underscores the country’s move toward more detailed guidance on how AI systems may interact with users and the wider economy.
Why this matters
The pattern across these updates is that AI regulation is becoming more operational and scenario-specific, focusing on automated decisions, synthetic sexual imagery, price-setting algorithms and conversational systems rather than abstract principles. For AI builders and enterprises, that means compliance is no longer just a matter of watching for headline AI acts; they must track how existing privacy, antitrust, sexual offences and sectoral statutes are being amended to capture AI behaviour. It also shows that regulators are trying to balance urgency with realism, introducing criminal offences and outright bans where harms are acute while offering grace periods and study mandates where the technology and its risks are still being mapped.
Looking ahead, the AI Law Radar changelog suggests that the next wave of AI law won’t arrive in one cinematic moment but through weekly adjustments, corrections and targeted additions across multiple jurisdictions. The UK’s tilt toward sandboxes over a standalone AI bill hints at a regulated experimentation model, while U.S. states like Illinois, New Jersey and Maryland are comfortable flatly prohibiting some algorithmic practices. As more entries move from proposed to effective and as confidence ratings rise on grace periods and enforcement details, AI companies can expect a regulatory landscape that is both more fragmented and more predictable, with the burden shifting from forecasting big-bang statutes to implementing concrete, domain-specific obligations in the products they ship.