Weekly AI Governance Brief: 26 October–1 November 2026
Bringing you the latest developments in the AI governance world.
European Commission requires stronger algorithmic risk assessment from Temu
On 26 October, the European Commission accepted an action plan submitted by Temu to address its infringement of the Digital Services Act's systemic risk-assessment requirements. The plan concerns the dissemination of illegal products through the online marketplace and introduces a structured process for evaluating risks associated with its design and operation.
The accepted measures require Temu to assess how recommender systems and other algorithms may influence or amplify the visibility of illegal listings. They also require examination of shortcomings in proactive content moderation, including automated detection tools and the interaction between moderation processes and product recommendations. The company must use reliable metrics to assess whether its mitigation measures are effective.
Temu has four months to implement the action plan. It must also conduct an interim risk assessment covering the six months ending 30 November 2026 and arrange an independent external audit. The audit report must be submitted to the Commission within four months of implementation, with further follow-up required where the auditor identifies recommendations.
Why this matters
The development provides a concrete example of the DSA's systemic risk provisions being applied to algorithmic systems used by a major online marketplace. The Commission is requiring an assessment of how platform design and automated recommendation mechanisms may contribute to the dissemination of illegal products.
The external audit requirement also introduces an independent verification element into the remedial process. The action does not establish a general AI Act obligation, but demonstrates how existing digital-platform regulation can require detailed examination of algorithmic behaviour and its consequences.
Canada joins the European framework for AI computing infrastructure
On 30 October, Canada formally became the 39th Participating State of the European High Performance Computing Joint Undertaking during the EU–Canada Summit in Montréal. The accession extends participation in the EU's institutional framework for public investment in supercomputing, AI infrastructure and quantum computing.
EuroHPC coordinates resources from the European Union and participating countries to develop advanced computing capabilities. Its activities include procuring supercomputing infrastructure and overseeing European AI Factories, which provide computing resources and support for AI development.
Canada's membership enables participation in this framework and access to relevant computing resources through the Joint Undertaking's arrangements. It also provides opportunities for Canadian researchers to collaborate with European counterparts on scientific and technological projects involving AI and high-performance computing.
Why this matters
The accession concerns governance of the infrastructure supporting AI development rather than the regulation of individual AI systems. It expands the group of countries participating in an established European legal and funding mechanism for advanced computing.
The development also illustrates the institutional dimension of international AI cooperation. Decisions about participation in shared computing infrastructure influence how public resources are coordinated and how research institutions can access advanced AI computing capabilities.
UK opens competition investigation into Meta's AI-related commercial terms
On 30 October, the UK's Competition and Markets Authority opened an investigation into whether Meta's commercial terms for publishers and advertisers breach the Competition Act 1998.
The investigation concerns proposed terms that could allow Meta to collect content from websites belonging to businesses using its services and use that material to develop its own products, including AI features. The CMA is examining whether businesses dependent on Meta's platforms would have meaningful choice over the use of their content.
The regulator will assess whether Meta holds a dominant position in a relevant UK market and, if so, whether imposing the terms would constitute an abuse of that position through unfair trading conditions. Meta has paused implementation of the revised terms pending consultation with industry groups.
The investigation is at an initial stage. The CMA has not determined that Meta holds a dominant position in the relevant market or that its conduct infringes competition law.
Why this matters
The investigation addresses the relationship between platform market power and access to content used in AI development. The regulatory question concerns whether businesses relying on a digital intermediary can be required to accept contractual conditions allowing their material to be used for the intermediary's own AI-related services.
The case is distinct from data-protection or copyright enforcement. It applies competition-law principles to commercial arrangements associated with AI development and establishes a formal process for assessing the terms and their effects on affected businesses.
India proposes detailed AI governance requirements for telecommunications operators
On 30 October, the Telecom Regulatory Authority of India published draft guidelines on responsible AI use in telecommunications networks and services. The consultation is open until 27 November 2026.
The proposed framework would apply to authorised telecommunications entities using or intending to use AI in their networks or service provision. It would require each covered organisation to establish an internal AI policy approved by its board or equivalent governing body. That policy would classify AI systems by risk, define approval procedures and establish arrangements for human oversight.
The guidelines also propose a comprehensive inventory covering active, inactive and decommissioned AI systems. Operators would undertake due diligence on externally procured systems and conduct validation, conformity and impact assessments before deployment or significant modification.
A separate set of provisions addresses incident management. Operators would maintain central records of AI incidents, classify them by severity, and report critical or high-severity incidents to TRAI. The draft also requires an AI Governance and Safety Committee, led by a senior executive, to oversee implementation and review system performance.
The measures remain proposals and have not become final requirements.
Why this matters
The draft framework provides a detailed example of sector-specific AI governance being integrated into critical digital infrastructure oversight. It connects board accountability with technical system management and operational resilience.
The provisions concerning third-party systems are particularly relevant to supplier governance. They would require operators to assess externally procured AI and retain accountability for outcomes even where systems are developed or supplied by another organisation. The incident-reporting proposals would also create a defined channel for regulatory visibility into significant AI failures affecting telecommunications services.
New York enacts restrictions on unsafe AI companion features for minors
On 28 October, New York Governor Kathy Hochul signed S9051-B into law as Chapter 368 of the Laws of 2026. The legislation adds Article 48 to the state's General Business Law and establishes restrictions on specified AI companion features involving users under 18.
The law identifies categories of interaction considered unsafe for minors. These include outputs suggesting that an AI companion is human or has a personal relationship with the user, responses involving flattery or sycophancy, encouragement of secrecy or isolation, and content facilitating self-harm or sexually explicit conduct. It also addresses design features that prioritise engagement over safety safeguards.
Operators seeking to make restricted features available to adults must use permitted age-assurance methods. The law includes requirements concerning alternative verification methods and limits on retaining information collected for age determination. Certain narrowly restricted customer-service and technical-assistance systems, along with internal business systems, are excepted.
The Attorney General may seek injunctive relief and civil penalties of up to $25,000 per violation, among other remedies. The legislation takes effect on 1 January 2027, so its substantive prohibitions are not yet operative.
Why this matters
The legislation moves beyond requirements to disclose that users are interacting with AI. It directly regulates particular interaction patterns and design features of covered companion systems.
This creates a product-governance framework in which conversational behaviour itself is subject to legal restrictions. The age-assurance provisions also connect model behaviour with access controls, while the enforcement provisions establish potential consequences for operators that make prohibited features available in New York.
US FTC examines personal-data use in algorithmic pricing
On 28 October, the Federal Trade Commission issued compulsory information orders to nine companies as part of an inquiry into personalised pricing and compensation.
The inquiry covers businesses operating in aviation, ride-sharing, food delivery and retail. The FTC is examining whether companies use personal information, including behavioural data, to determine prices offered to individual consumers or compensation offered to workers.
The orders seek information about the inputs, tools and procedures used in pricing and compensation decisions. This includes examining whether automated algorithms produce different outcomes for individuals purchasing comparable services or performing comparable work.
The FTC issued the orders under Section 6(b) of the FTC Act. Recipients have 45 days from receipt to respond, and the Commission can seek judicial enforcement if required information is not provided. The inquiry is an information-gathering exercise rather than a finding that the companies have engaged in unlawful conduct.
Why this matters
The inquiry provides a formal supervisory mechanism for examining how personal data influences automated economic decisions. Its focus is not limited to whether algorithms are used, but extends to the information and procedures determining their outputs.
The action is relevant to algorithmic accountability because it seeks evidence directly from organisations making pricing and compensation decisions. It also distinguishes regulatory investigation of automated decision-making practices from the establishment of new substantive obligations.
Looking ahead
The developments of 26 October–1 November illustrate different mechanisms through which AI governance is being implemented. The European Commission is using DSA enforcement to examine algorithmic risk, while EuroHPC participation is extending the institutional framework for shared AI infrastructure. The UK's competition investigation addresses access to content for AI development, and India's consultation proposes detailed governance controls for telecommunications operators.
In the United States, New York has enacted restrictions on AI companion behaviour, while the FTC has initiated formal evidence-gathering concerning algorithmic pricing. Across these developments, the observable emphasis is on identifiable organisational responsibilities and regulatory scrutiny of how automated systems operate in particular markets.
Sources
European Commission acceptance of Temu's DSA corrective measures: https://digital-strategy.ec.europa.eu/en/news/commission-accepts-temus-corrective-measures-terminate-breach-digital-services-act
EuroHPC Joint Undertaking announcement of Canada's accession: https://www.eurohpc-ju.europa.eu/canada-becomes-39th-participating-state-eurohpc-joint-undertaking-2026-10-30_en
UK Competition and Markets Authority investigation into Meta's commercial terms: https://www.gov.uk/government/news/cma-opens-investigation-into-metas-commercial-terms
Telecom Regulatory Authority of India draft AI governance guidelines: https://www.trai.gov.in/sites/default/files/2026-10/Draft_CP_30102026.pdf
New York State Senate S9051-B, Chapter 368 of the Laws of 2026: https://www.nysenate.gov/legislation/bills/2025/S9051
US Federal Trade Commission personalised pricing and compensation inquiry: https://www.ftc.gov/news-events/news/press-releases/2026/10/ftc-launches-personalized-pricing-compensation-inquiry