By As Law Online 
California has long set the pace for U.S. technology policy, and recent moves show the state is not slowing down. In 2025 alone, lawmakers and the governor advanced rules that reach deep into how the most powerful AI systems are built and how large tech deals get reviewed. For tech executives, AI developers, corporate counsel, compliance officers, startup founders, policy analysts, and venture capitalists, understanding these shifts is no longer optional. This article explains why California targets AI safety & big mergers, breaks down the core requirements of Senate Bill 53 (the Transparency in Frontier Artificial Intelligence Act, or TFAIA), examines the parallel push on antitrust and merger scrutiny, and offers practical steps for compliance and strategic adaptation.
The Policy Drivers Behind California’s Dual Focus
California is home to a large share of the world’s leading AI labs and the densest concentration of venture-backed tech startups. That concentration creates both economic opportunity and political pressure. Lawmakers see frontier AI systems as potential sources of catastrophic risk (chemical, biological, radiological, or nuclear weapons assistance, large-scale cyber attacks without meaningful human oversight, or models that evade developer control). At the same time, they view serial acquisitions and concentrated market power as threats to the competitive environment that once allowed new entrants to thrive.
Governor Gavin Newsom’s office has repeatedly framed the state’s approach as a balance: protect public safety and competition without choking innovation. After vetoing the broader SB 1047 in 2024, the administration supported a narrower transparency-focused bill in 2025. Parallel work by the California Law Revision Commission and Attorney General Rob Bonta has kept merger and monopoly issues on the agenda. The result is a two-track strategy: transparency and incident reporting for the largest AI model developers, and expanded tools for reviewing and challenging big tech deals.
Understanding Senate Bill 53 and the Transparency in Frontier Artificial Intelligence Act
SB 53, signed by Governor Newsom on September 29, 2025, and chaptered as Chapter 138 of the Statutes of 2025, makes California the first state to enact a statute specifically aimed at frontier AI safety and transparency. The law is commonly called the Transparency in Frontier Artificial Intelligence Act (TFAIA).
Key Definitions That Determine Scope
- A frontier model is a foundation model trained using a quantity of computing power greater than 1026 integer or floating-point operations. The threshold includes the original training run plus subsequent fine-tuning, reinforcement learning, or other material modifications.
- A frontier developer is any person who has trained or initiated the training of a frontier model.
- A large frontier developer is a frontier developer (together with its affiliates) that had annual gross revenues exceeding $500 million in the preceding calendar year.
These thresholds are designed to reach only the most capable systems and the best-resourced developers. In practice, the large-developer category currently covers a small number of major labs. The Department of Technology is directed to assess the definitions periodically and recommend updates so they stay aligned with technological developments and national or international standards.
Catastrophic Risk Framework
The statute centers on catastrophic risk, defined as a foreseeable and material risk that a frontier model’s development, storage, use, or deployment will materially contribute to the death of or serious injury to more than 50 people, or more than $1 billion in damage to or loss of property, arising from a single incident involving:
- AI-assisted creation or release of a chemical, biological, radiological, or nuclear weapon,
- cyber attacks or serious crimes caused by an AI system without meaningful human oversight, or
- activity that evades the control of its developer or user.
This definition keeps the focus on high-severity, low-probability scenarios rather than everyday harms such as bias or misinformation.
Core Obligations for Large Frontier Developers
Large frontier developers must write, implement, comply with, and clearly and conspicuously publish a frontier AI framework. The framework must describe how the developer approaches:
- incorporation of national standards, international standards, and industry-consensus best practices,
- identification, assessment, and mitigation of catastrophic risks,
- internal governance structures,
- cybersecurity practices to secure unreleased model weights,
- assessment and management of catastrophic risk from internal use of models (including risks that a model could circumvent human oversight),
- processes for responding to critical safety incidents, and
- use of third parties where appropriate for independent assessment.
The framework must be updated at least annually and within 30 days of any material modification. Redactions for trade secrets or cybersecurity are permitted under defined conditions.
Transparency Reports and Incident Reporting
All frontier developers (not only the large ones) must publish a transparency report before or concurrently with deploying a new frontier model or a substantially modified version of an existing one. The report must include, among other items:
- a mechanism for individuals to communicate with the developer,
- the release date,
- supported modalities and languages,
- intended uses and any restrictions or conditions,
- and, for large frontier developers, a summary of the frontier AI framework and related risk assessments.
Critical safety incidents must be reported to the California Office of Emergency Services (Cal OES). The standard window is 15 days after discovery; if the incident poses an imminent threat of death or serious physical injury, the window shrinks to 24 hours. Cal OES must establish mechanisms for both developers and members of the public to report incidents, and large frontier developers must also be able to submit confidential summaries of internal catastrophic-risk assessments.
Whistleblower Protections
The law adds robust AI whistleblower protections under the Labor Code. Employees who report potential catastrophic risks or violations of the TFAIA receive protection from retaliation. Successful plaintiffs may recover attorneys’ fees. These provisions create internal accountability channels that regulators and the public can rely on when external visibility is limited.
Enforcement and Penalties
The California Attorney General enforces the TFAIA through civil penalties of up to $1 million per violation. The statute also creates a pathway for a consortium to develop CalCompute, a public cloud computing platform intended to expand safe and equitable AI research access.
Why California Is Also Heightening Scrutiny of Big Mergers
While SB 53 targets model development, California has simultaneously expanded its toolkit for reviewing and challenging large technology transactions. The state’s approach rests on three pillars: better visibility into proposed deals, potential changes to substantive merger standards, and stronger single-firm conduct rules.
Pre-Merger Notification Requirements
In early 2026 California enacted the California Uniform Antitrust Pre-Merger Notification Act (often referred to as SB 25). Beginning January 1, 2027, parties that file a federal Hart-Scott-Rodino (HSR) notification and meet certain California-nexus thresholds must also provide a copy of the filing materials to the California Attorney General within one business day. The requirement is non-suspensory (it does not create a separate waiting period), but it gives state enforcers earlier and more complete information. California joins other states that have adopted similar “mini-HSR” regimes, creating a growing compliance patchwork for multi-state deals.
California Law Revision Commission Merger Proposals
The California Law Revision Commission (CLRC), tasked by the Legislature with studying updates to the Cartwright Act, has examined options for a state-level merger control regime. Staff proposals have included:
- tracking or expanding the Clayton Act standard while explicitly covering monopsony,
- creating rebuttable presumptions based on market share (around 30 percent) or Herfindahl-Hirschman Index (HHI) thresholds,
- recognizing the 2023 Federal Merger Guidelines as persuasive authority, and
- exploring an “appreciable risk” standard that would prohibit mergers creating even a meaningful chance of reducing competition.
These options remain under consideration. Industry groups have warned that rigid structural presumptions or novel standards could chill the acquisition pathway that many startups rely on for exits and scale, particularly in capital-intensive fields such as AI.
Expanding the Cartwright Act: The COMPETE Act and Related Bills
Assembly Bill 1776, known as the COMPETE Act, would amend the Cartwright Act to prohibit single-firm monopolization and monopsonization. Current California law largely addresses coordinated conduct among multiple firms. AB 1776 would close that gap, declare that federal antitrust interpretations are “at most instructive, and not conclusive,” and instruct courts to liberally interpret the state’s laws to promote free and fair competition. Small businesses meeting defined thresholds would be exempt. As of mid-2026 the bill had passed the Assembly and cleared the Senate Judiciary Committee with amendments; its final fate remains subject to further legislative action and the governor’s signature.
Other measures have targeted self-preferencing by large platforms and algorithmic pricing. AB 325, signed in 2025, makes it unlawful under the Cartwright Act to use or distribute a common pricing algorithm as part of a contract or conspiracy to restrain trade, and lowers certain pleading barriers for plaintiffs.
Active Enforcement by the Attorney General
Attorney General Rob Bonta has used existing federal and state tools aggressively. In 2026 his office led a multi-state coalition challenging the proposed Paramount Skydance–Warner Bros. Discovery merger on Clayton Act grounds, illustrating California’s willingness to step in even when federal agencies clear a deal. The office has also increased resources for antitrust work and issued advisories reminding businesses that California is not a regulatory vacuum.
Taken together, these developments mean that Silicon Valley deal scrutiny is rising. Acqui-hires, talent-and-IP deals structured to avoid formal merger filings, and serial acquisitions in AI-adjacent markets face closer examination at both federal and state levels.
Impact on Frontier Model Developers and Compliance Requirements
For companies that meet the frontier-model or large-developer thresholds, SB 53 converts many voluntary safety practices into legal duties. The operational burden falls heaviest on governance, documentation, and reporting systems.
Developers must map existing internal safety processes to the statutory frontier AI framework elements. Gaps in third-party assessment, cybersecurity of model weights, or internal-use risk evaluation need prompt attention. Transparency reports must be ready before public deployment of new or substantially modified models. Incident-response playbooks must incorporate Cal OES timelines and confidentiality options for internal assessments.
Whistleblower programs require review to ensure they cover the specific catastrophic-risk and TFAIA violation categories and that employees know how to use them without fear of retaliation. Legal and compliance teams should coordinate with product and research leadership so that risk assessments feed both internal decision-making and the required public and confidential disclosures.
Penalties of up to $1 million per violation create real financial exposure. More importantly, public frameworks and transparency reports will become reference points for customers, investors, partners, and future regulators. Inaccurate or incomplete disclosures can generate reputational and litigation risk beyond the statutory fines.
Impact of Heightened Merger Scrutiny on Startups, Acquirers, and Investors
Startup founders and venture capitalists face a more complex exit environment. Large strategic acquirers must now plan for earlier state notification, longer diligence timelines, and the possibility of state challenges even after federal clearance. Deals involving AI talent, proprietary models, or data assets attract particular attention because they can eliminate nascent competitive threats.
Investors should factor regulatory uncertainty into term sheets and board discussions. Acqui-hire structures that once avoided HSR filings may still draw scrutiny if they appear designed to evade review. Cross-market efficiencies that are common in platform and AI deals may receive less weight under some of the CLRC proposals or under a strengthened Cartwright Act that limits offsetting benefits across markets.
For corporate development teams, the practical response is earlier antitrust counseling, more detailed competitive-effects analysis, and contingency planning for remedies or litigation. California’s size and the national reach of its major tech firms mean that state standards can influence national deal practice.
Practical Compliance and Strategic Adaptation Guide
Tech leaders can take concrete steps now.
For frontier AI developers
- Inventory models against the 1026 FLOPs threshold, including cumulative post-training compute.
- Draft or update the frontier AI framework to cover every required element and schedule annual and event-driven reviews.
- Establish clear ownership for transparency-report production and Cal OES reporting.
- Strengthen whistleblower channels and train relevant employees.
- Engage outside counsel experienced in both AI safety regulation and California enforcement practice.
- Monitor Department of Technology recommendations on definitional updates.
For companies considering or defending large transactions
- Build California notification into deal checklists once SB 25 takes effect.
- Conduct competitive analysis that anticipates both federal and potential state standards, including monopsony and structural presumptions.
- Document efficiencies and innovation benefits carefully, recognizing that some proposed reforms would limit cross-market offsetting.
- Evaluate alternative structures (licensing, collaboration agreements, minority investments) with antitrust counsel if a full acquisition appears high-risk.
- Track AB 1776 and CLRC merger recommendations through the remainder of the 2025–2026 legislative session and beyond.
Cross-cutting recommendations
- Treat California rules as a baseline that may influence other states and any eventual federal framework.
- Align internal AI governance and M&A processes so that safety and competition considerations are considered together rather than in silos.
- Maintain open channels with policymakers; the iterative nature of California AI legislation (SB 1047 veto followed by SB 53 passage) shows that engagement can shape outcomes.
- Budget for increased legal, compliance, and external-assessment costs as a normal cost of operating at the frontier.
Industry Reaction and the Broader Policy Landscape
Reactions have been mixed. Anthropic publicly endorsed SB 53, arguing that transparency requirements formalize practices many labs already follow. Other developers preferred federal harmonization and expressed concern about a patchwork of state rules. Venture groups have warned that stricter merger standards could reduce exit options and therefore investment in early-stage AI companies. Supporters of the legislation, including safety-focused advocacy organizations, view the laws as a necessary “trust but verify” approach that keeps California competitive while addressing genuine catastrophic risks.
At the federal level, bipartisan proposals such as the FRONTIER Act have appeared, some of which would require similar transparency and incident reporting while potentially preempting conflicting state rules. Until federal legislation is enacted, California’s framework remains the leading U.S. model for frontier AI governance.
Looking Ahead: Adaptation as a Competitive Advantage
California targets AI safety & big mergers because the state sits at the center of both the technology and the political debate. SB 53 and the related antitrust measures do not ban development or acquisitions; they impose transparency, accountability, and heightened review. Companies that treat these requirements as strategic design constraints rather than pure compliance burdens will be better positioned.
Developers that can point to robust, published frameworks and clean incident histories will enjoy stronger relationships with enterprise customers and regulators. Acquirers that anticipate state-level scrutiny and structure deals accordingly will close more reliably. Startups that understand the new exit calculus can plan capital strategies and partnership options with clearer eyes.
The legal landscape will continue to evolve. Definitional thresholds may be updated, additional states may follow California’s lead, and federal action remains possible. The practical response is the same: build durable governance systems, document decisions carefully, and stay engaged with the policy process.
Tech executives, developers, counsel, and investors who master these requirements will not only reduce legal risk; they will help shape the standards that govern the next generation of AI systems and the competitive structure of the industry that builds them. Consult experienced California counsel early, map your specific exposure, and treat compliance as an ongoing operational discipline rather than a one-time project. The companies that do so will navigate the current wave of regulation with greater confidence and continue to innovate inside the guardrails California has set.