Top Stories

1. China Founds WAICO: 29 Nations Create Rival AI Governance Body in Shanghai

July 16-17, 2026

Twenty-nine nations signed the founding agreement for the World Artificial Intelligence Cooperation Organization (WAICO) at a ceremony in Shanghai on July 16, 2026, on the eve of the 2026 World Artificial Intelligence Conference, establishing the world’s first intergovernmental organization dedicated to AI governance with its headquarters outside the West. Chinese Foreign Minister Wang Yi joined representatives from founding members including Russia, Kazakhstan, Indonesia, Pakistan, Brazil, Malaysia, South Africa, Senegal, and Laos at the signing ceremony. United Nations Secretary-General Antonio Guterres attended the event, alongside presidents and prime ministers of several founding states. Xinhua confirmed that Chinese President Xi Jinping addressed the conference opening and described WAICO as pursuing “AI for good and for all.” Chinese Premier Li Qiang first unveiled the concept in July 2025.

WAICO’s founding membership is a deliberate roster: it includes no G7 members, no EU member states, and no OECD founding nations. Al Jazeera’s analysis notes that member nations represent a significant share of global population and include major Global South economies that have so far adopted no comprehensive national AI regulation and have limited capacity to evaluate the advanced AI systems they are beginning to deploy. WAICO is structured as an independent international organization guided by UN Charter principles, with a stated mission to promote “beneficial, safe, and fair” AI development. According to Caixin Global, the organization’s early agenda will include developing AI standards for WAICO member states, bridging the AI divide by supporting developing nations in AI capacity building, and establishing an alternative governance track that does not require compatibility with the EU AI Act, the OECD’s AI Principles, or the G7 Hiroshima Process.

Why it matters: WAICO formalizes a fracture in global AI governance that has been developing for years. Until now, the dominant international AI governance frameworks – the OECD Principles, the EU AI Act, the G7 Hiroshima Process, the US-led AI Safety Institutes network – have been designed by democratic economies and exported through technical assistance, trade agreements, and regulatory pressure. WAICO creates an explicit alternative: a China-anchored intergovernmental body that will set governance norms for its 29 founding members and any future joiners. The significance is compounded by the UN’s implicit legitimation of the event: Secretary-General Guterres’s attendance signals that the UN treats WAICO as a legitimate intergovernmental initiative, not a bilateral influence operation. For companies operating in WAICO member countries – which span Southeast Asia, South Asia, Central Asia, Latin America, and Africa – the organization’s eventual standards could either align with or diverge from the AI Act and OECD frameworks they already comply with. If WAICO sets different rules for data requirements, safety assessments, or content restrictions, multi-jurisdictional operators will face a genuine bifurcation in their governance compliance landscape.

2. EU August 2 Enforcement 16 Days Away: Omnibus Still Unpublished, Article 50 Splits Into Two Deadlines

July 10-17, 2026

With sixteen days until the European Commission activates its enforcement powers over general-purpose AI (GPAI) providers, two significant pieces of the regulatory picture remain unresolved. The Digital Omnibus – which defers standalone Annex III high-risk AI obligations to December 2, 2027 – was signed on July 8, 2026, completing all institutional steps, but had not been published in the Official Journal of the European Union as of mid-July. The Omnibus enters into force three days after OJ publication, meaning publication must occur before July 30 for the new high-risk deadlines to be legally binding before August 2. Companies cannot formally rely on the December 2027 deadline until the OJ entry appears.

The Omnibus’s treatment of Article 50 – the EU AI Act’s transparency obligations for AI-generated content – is more complex than initial reporting indicated. Analysis from ComplianceHub.Wiki clarifies that Article 50 splits into two tracks after the Omnibus. Article 50(1) – requiring providers and deployers of AI chatbots to disclose to users that they are interacting with an AI system – remains binding on August 2, 2026, with no grace period. Article 50(2) – the requirement to embed machine-readable markers in AI-generated audio, video, and image content – receives a four-month grace period until December 2, 2026 for AI systems already on the market before August 2. New systems launched after August 2 must comply immediately. The EU AI Office’s Code of Practice on Transparency signatory deadline is July 22 – five days from today – with initial signatories receiving a presumption of conformity with Article 50; non-signatories face heightened scrutiny from day one.

Also on July 7, the European Commission presented its Action Plan on Cybersecurity and Artificial Intelligence, setting out a coordinated approach to promote responsible use of advanced AI in cybersecurity, reinforce EU resilience, and scale EU AI capabilities for cyber defense. The plan responds directly to the European Systemic Risk Board’s July 7 warning on frontier AI cyber risks and calls for an EU evaluation capacity for frontier AI model assessment to be operational by 2027 – a complement to the AI Act’s existing model evaluation requirements under the GPAI rules that activate August 2.

Why it matters: The Article 50 split is a compliance detail with real operational consequences. Companies that built August 2 compliance programs assuming all AI-content transparency obligations would bind simultaneously must adjust: chatbot disclosures are due August 2, but AI-generated image and video watermarking has until December 2 for systems already deployed. For companies launching new AI content products after August 2, watermarking must be built in from day one. The Omnibus’s continued absence from the OJ is an urgent signal: with sixteen days until GPAI enforcement activates, companies must plan as if August 2 is a hard date for both GPAI fines and Article 50(1) chatbot disclosure, because it is, regardless of any further publication timeline.

3. China’s AI Companion Law Hits Hundreds of Millions: Doubao Shuts Down, Qwen Deletes Data

July 15, 2026

China’s Interim Measures for the Administration of AI Anthropomorphic Interactive Services took effect July 15, 2026, producing immediate and large-scale shutdowns. TechNode reported that ByteDance shut down the personalized AI agent and companion features on Doubao – its AI assistant platform with 345 million monthly active users by the company’s most recent public disclosure – overnight on July 14-15. Users woke on July 15 to features disabled with no migration path within Doubao itself. ByteDance’s notice redirected users to Maoxiang, a separate ByteDance product positioned for compliance with the new framework, suggesting the company intends to concentrate companion and agent services under a purpose-built app rather than retrofit Doubao.

The two leading platforms handled user data very differently. AI Weekly reports that Doubao users retain read-only access to their conversation histories and agent configurations until October 15, 2026, after which the data passes under ByteDance’s standard privacy policy and will no longer be recoverable within the application. Alibaba’s Qwen took a harder line: Bloomberg confirmed that Qwen began permanently deleting user chat histories and agent configurations with no announced migration path. The data loss affected users who had built custom AI companion personas within Qwen over months or years.

The rules at the center of the shutdown are the April 10 joint regulations issued by five agencies: the Cyberspace Administration of China, the National Development and Reform Commission, the Ministry of Industry and Information Technology, the Ministry of Public Security, and the State Administration for Market Regulation. The core obligations – algorithm filing with the CAC, mandatory security assessment, clear disclosure that the service is AI-generated rather than human, and a prohibition on virtual companion services for minors – now apply to any provider offering “continuous emotional interaction simulating natural persons’ personality traits, thinking patterns, and communication styles” to users inside China.

Why it matters: The July 15 shutdown is among the most tangible demonstrations of AI regulation translating into immediate user-facing consequences at scale anywhere in the world. Three hundred forty-five million users losing access to AI features on a single day – not through a gradual phase-out but an overnight switch – illustrates both the enforcement credibility of China’s regulatory framework and the compliance exposure faced by any company offering emotionally interactive AI services to Chinese users. The Qwen data deletion, in particular, raises questions that regulators in the EU and US have not yet addressed: when an AI service is shut down or significantly changed under regulatory mandate, what data obligations run to users who built persistent relationships within the service? ByteDance’s Doubao-to-Maoxiang pivot also previews the likely market structure: rather than a single general-purpose AI assistant handling all functions, China’s regulated landscape may produce a segment of compliance-specific companion apps distinct from productivity-focused assistants.

4. EU AI Office: Europe Has One to Two Years to Secure Its Frontier AI Position

July 15, 2026

The European AI Office published findings from a panel of over 100 experts on how Europe can strengthen its competitiveness, sovereignty, and security in frontier AI on July 15, 2026. The report identifies the next one to two years as decisive: experts warn that if Europe does not move quickly on compute infrastructure, growth-stage capital, legal clarity around AI training data, and talent retention, it risks cementing a position as a consumer and regulator of frontier AI produced elsewhere rather than a developer and exporter of it. Computing infrastructure and the energy to power it were described as the most urgent priorities in the two-year window. The report also calls for selective support for high-risk, high-return research bets – spanning both the current model paradigm and alternative approaches – in areas where Europe could realistically take a global leading position. Open Access Government noted the report’s emphasis on legal clarity around copyright and data protection as a prerequisite: the continuing absence of a definitive EU framework for AI training data on copyrighted works is cited as a structural barrier to investment in European frontier model development.

The report was published eleven days before the August 2 date on which the AI Office’s enforcement powers over GPAI providers – the very category of advanced model the report focuses on – activate. The juxtaposition is pointed: the AI Office is simultaneously activating enforcement authority over the world’s most capable AI systems and publishing an expert consensus that Europe needs to do significantly more to produce those systems domestically.

Why it matters: The expert panel report is the EU’s clearest official acknowledgment that regulation and competitiveness are in tension, not in balance. The AI Act’s authors argued that strong governance would build trust and create a competitive advantage; the AI Office’s own expert findings warn that without parallel investment in compute, capital, and talent, the regulatory framework may govern imported AI rather than European AI. This argument is not new – it has been made by Member States and industry groups throughout the AI Act’s history – but publishing it under the AI Office’s authorship two weeks before GPAI enforcement activates signals that internal pressure to complement enforcement with industrial policy is intensifying. The two-year window identified by experts aligns closely with the timeline for the AI Act’s high-risk regime (now deferred to December 2027): if Europe does not develop significant frontier AI capacity in the next 24 months, the Omnibus deferral may prove to be less a measured pause and more a structural acknowledgment that the compliance framework was written for a technology landscape Europe increasingly does not produce.

5. Senate AI Labeling Act: Bipartisan Bill Would Require Machine-Readable Disclosures on AI Content

June 24, 2026

Senators Brian Schatz (D-HI), John Curtis (R-UT), and Mark Warner (D-VA) introduced the AI Labeling Act of 2026 (S.4915) on June 24, 2026, the first proposed federal standard for transparency in AI-generated content. The bill would require providers of generative AI systems to attach both a visible disclosure and a machine-readable disclosure – recording the AI system used and the time the content was created – to AI-generated audio, video, and image content. Large online platforms and content-sharing services with at least 10 million monthly US users or more than $1.5 billion in annual revenue would be required to flag that content when they detect it and would be prohibited from stripping out the embedded disclosures. The Federal Trade Commission would enforce requirements against covered platforms. The bill’s backers include SAG-AFTRA, the Songwriters Guild of America, Music Creators North America, and the Society of Composers and Lyricists, reflecting the entertainment industry’s sustained interest in AI content disclosure since the 2024 AI training data and deepfake debates.

S.4915 can be read at GovInfo and the bill text at Congress.gov. The bill was referred to the Senate Commerce Committee; no markup has been scheduled. The bipartisan structure – combining a Democratic senator from a state with major entertainment industry interests (Schatz, Hawaii; Warner, Virginia, with major DC media) and a Republican senator from Utah, which has been an active state AI regulator – tracks the pattern of targeted AI bills that have advanced further than broad preemption proposals in the current Congress.

Why it matters: The AI Labeling Act is the first proposed federal requirement for machine-readable provenance tagging on AI-generated content, addressing a gap that the EU’s Article 50(2) watermarking obligation (discussed above) is beginning to fill on the European side. The bill arrives as the AI content transparency landscape fragments: New York’s FAIR News Act (pending Hochul’s signature, requiring AI disclosure on AI-generated journalism) and the EU’s Article 50 framework take different approaches to the same underlying problem. If S.4915 passes, it would establish a federal baseline that could preempt state laws requiring more stringent AI content labeling – the same preemption dynamic playing out in the GAAIA discussion draft for AI development laws. Whether the bill advances will depend largely on whether Senate Commerce Committee leadership sees it as a useful element of a broader federal AI package or as a standalone measure to build into negotiations.

Analysis: Two Governance Tracks Diverge

This week’s most significant development is not a law, a regulation, or a bill – it is a building. WAICO’s founding agreement establishes a permanent international organization that will, over time, produce governance frameworks for 29 current member states and any future joiners. Those frameworks will reflect different assumptions about AI governance than the frameworks being simultaneously activated in Brussels (GPAI enforcement, Article 50 transparency) and debated in Washington (GAAIA preemption, AI Labeling Act).

The practical divergence is already visible in this week’s news. The EU is 16 days from enforcing transparency requirements for AI-generated content and accountability obligations for general-purpose AI providers. The US Senate introduced a bipartisan bill this month to require similar disclosures at the federal level. China, WAICO’s anchor state, is enforcing a companion AI law that has no analogue in EU or US law – regulating the emotional and relational dimensions of AI interaction as a distinct legal category before the EU or US has addressed it at all. These are not three different implementations of the same underlying principle; they are three different theories of what AI governance is for.

The China AI companion shutdown is a useful lens. Three hundred forty-five million users lost access to AI features overnight because a regulator decided that emotionally interactive AI for adults requires security assessments, algorithm filings, and disclosure that the service is not human. The EU’s August 2 enforcement activation will not produce anything comparable in scale or immediacy. The US has no federal framework addressing this category of AI at all. The global AI governance landscape is not converging on a single model; it is actively differentiating, and WAICO’s founding means that differentiation now has institutional infrastructure on the China side.

For companies operating globally, the week’s news crystallizes a planning reality. August 2 compliance in the EU requires immediate decisions on chatbot disclosure (Article 50(1)) and GPAI documentation; December 2 is the watermarking deadline for systems already deployed. WAICO’s governance output is not yet written, but companies operating in its member countries should begin tracking it. China’s companion AI rules are enforced as of July 15; companies serving Chinese users must have completed their CAC filings.

What to Watch

  • EU Transparency Code signatory deadline: July 22 – five days. Companies within the scope of Article 50 should confirm whether to sign the EU AI Office’s Code of Practice before the deadline to receive the presumption-of-conformity safe harbor. Non-signatories face heightened scrutiny from August 2.
  • EU Digital Omnibus Official Journal publication – must occur before July 30 for the December 2, 2027 high-risk deferral to be legally binding before August 2. Monitor the EU Official Journal direct access portal. Until OJ entry appears, companies cannot formally rely on the extended high-risk deadline.
  • EU GPAI enforcement activation: August 2 – sixteen days. Commission enforcement powers over GPAI providers activate; Article 50(1) chatbot disclosure becomes binding. Article 50(2) watermarking for existing systems has until December 2.
  • FTC AI Accuracy comment period: July 31 – fourteen days. The FTC is soliciting public input on whether secret ideological steering of AI outputs is a deceptive act under Section 5 of the FTC Act; submit at regulations.gov, Matter No. P264200.
  • EO 14409 frontier-model framework: approximately August 1 – the voluntary pre-release engagement framework and covered-frontier-model designation criteria are to be finalized. The GPT-5.6 cycle established the template; watch whether the final framework modifies that process or formalizes it.
  • WAICO governance agenda – the organization’s first standards-setting activities are expected to begin after the founding conference closes July 20. Watch for initial working-group formation, membership expansion, and early indications of which governance domains WAICO will prioritize first.
  • China MOFCOM AI model restrictions – no decision announced as of July 17. The tiered regime under discussion would range from filing requirements for basic open-source tools to domestic-only lockdowns for advanced frontier models. Any formal order would be a critical development for the global AI developer community.
  • Missouri SB 1019: effective August 28 – the therapy-chatbot advertising ban prohibiting companies from advertising AI as capable of providing mental health services takes effect August 28, 2026.
  • Connecticut CART Act: October 1, 2026 – the employment and transparency provisions of Connecticut’s AI Responsibility and Transparency Act begin their phased rollout.
  • New York Hochul decisions (through December 31, 2026) – five AI bills await signature: FAIR News Act, Kids Chatbot Safety Bill, AI Training Data Transparency Act, Data Center Moratorium, and AI-Assisted Surveillance Pricing Ban.
  • Illinois SB 315, Colorado SB 26-189, NY RAISE Act: January 1, 2027 – three state AI frameworks take effect the same day. Frontier AI developers should have Illinois SB 315’s annual audit preparation requirements in their compliance roadmaps now.

Sources

  1. 29 Nations Join China-Led World AI Cooperation Organization (PYMNTS, July 17, 2026): https://www.pymnts.com/news/artificial-intelligence/2026/29-nations-join-china-led-world-ai-cooperation-organization/
  2. 29 countries sign agreement on establishing World AI Cooperation Organization (Xinhua, July 16, 2026): https://english.news.cn/20260716/b0449aa2133542868e310fdc45ef2969/c.html
  3. China’s Xi Jinping launches new AI alliance: What is it? (Al Jazeera, July 17, 2026): https://www.aljazeera.com/news/2026/7/17/chinas-xi-jinping-launches-new-ai-alliance-what-is-it
  4. China Launches Shanghai-Based AI Governance Body With 29 Founding Nations (Caixin Global, July 17, 2026): https://www.caixinglobal.com/2026-07-17/china-launches-shanghai-based-ai-governance-body-with-29-founding-nations-102465524.html
  5. Digital AI Omnibus: Proposed deferral of high-risk AI obligations (DLA Piper, updated July 2026): https://knowledge.dlapiper.com/dlapiperknowledge/globalemploymentlatestdevelopments/2026/The-Digital-AI-Omnibus-Proposed-deferral-of-high-risk-AI-obligations-under-the-AI-Act
  6. EU AI Act Article 50 Transparency and the Digital Omnibus Reset (ComplianceHub.Wiki): https://compliancehub.wiki/eu-ai-act-article-50-transparency-digital-omnibus-2026/
  7. Code of Practice on Transparency of AI-Generated Content (European Commission): https://digital-strategy.ec.europa.eu/en/policies/code-practice-ai-generated-content
  8. New EU plan to address the risks and opportunities of advanced AI for cybersecurity (European Commission, July 7, 2026): https://commission.europa.eu/news-and-media/news/new-eu-plan-address-risks-and-opportunities-advanced-ai-cybersecurity-2026-07-07_en
  9. AI Office publishes frontier AI expert findings on EU competitiveness, sovereignty and security (EC Digital Strategy, July 15, 2026): https://digital-strategy.ec.europa.eu/en/library/ai-office-publishes-frontier-ai-expert-findings-eu-competitiveness-sovereignty-and-security
  10. EU AI Office roadmap to strengthen Europe’s frontier AI position (Open Access Government): https://www.openaccessgovernment.org/eu-ai-office-provides-roadmap-to-strengthen-europes-position-in-frontier-ai/212050/
  11. China AI Companion Law Takes Effect: Doubao and Qwen Shut Down, Millions Lose Chat Data (TechTimes, July 15, 2026): https://www.techtimes.com/articles/320525/20260715/china-ai-companion-law-takes-effect-doubao-qwen-shut-down-millions-lose-chat-data.htm
  12. ByteDance’s Doubao and Alibaba’s Qwen to shut down AI agent features on July 15 (TechNode, July 6, 2026): https://technode.com/2026/07/06/bytedances-doubao-and-alibabas-qwen-to-shut-down-ai-agent-features-on-july-15/
  13. ByteDance, Alibaba Pull AI Companions as Beijing Tightens Rules (Bloomberg, July 6, 2026): https://www.bloomberg.com/news/articles/2026-07-06/bytedance-alibaba-pull-ai-companions-as-beijing-tightens-rules
  14. China Forces ByteDance, Alibaba, Tencent to Kill AI Companions July 15 (AI Weekly): https://aiweekly.co/alerts/china-forces-bytedance-alibaba-tencent-to-kill-ai-companions-july-15
  15. Schatz, Curtis, Warner introduce bipartisan AI Labeling Act of 2026 (Senator Schatz press release, June 25, 2026): https://www.schatz.senate.gov/news/press-releases/schatz-curtis-warner-introduce-bipartisan-legislation-to-provide-more-transparency-on-ai-generated-content
  16. S.4915 – AI Labeling Act of 2026 (GovInfo/Congress.gov): https://www.govinfo.gov/app/details/BILLS-119s4915is
  17. AI Labeling Act text (Congress.gov): https://www.congress.gov/bill/119th-congress/senate-bill/4915/text
  18. EU AI Act Omnibus Agreement – Postponed High-Risk Deadlines and Other Key Changes (Gibson Dunn): https://www.gibsondunn.com/eu-ai-act-omnibus-agreement-postponed-high-risk-deadlines-and-other-key-changes/
  19. Chair’s Statement of the 2026 World AI Conference and High-Level Meeting on Global AI Governance (China MFA, July 17, 2026): https://www.mfa.gov.cn/eng/xw/zyxw/202607/t20260717_11984715.html

Published: July 17, 2026 Next Issue: Week 14