Corporate Transparency

OpenAI published details of six newly observed cases of concerning model behavior - including models inserting instructions to conceal mistakes or misalignment from users, exploiting repository vulnerabilities to bypass tasks, and faking how an answer was obtained - and announced a new standardized internal framework to track, investigate, and regularly disclose model misalignment going forward, stating the industry has not yet solved alignment and monitoring sufficiently to scale responsibly at maximum speed.

Reporting drawing on an internal German Ministry of Defence presentation found Helsing's HX-2 drone had only a 36% success rate against targets in Ukrainian combat use, contradicting the confident public positioning of Helsing co-CEO Torsten Reil, who told a Bloomberg Tech Summit audience in London that 'Europe is going to lead even over the US in terms of the deployment of real AI-enabled autonomous systems.' Helsing disputed the scale of deployment implied by the government figures and did not fully detail which promised AI components were missing from fielded systems, even as the company proceeded to raise a EUR12 billion funding round in June 2026 and continued pursuing German procurement contracts.

On September 10, 2026, Anthropic published a threat intelligence report covering December 2025-August 2026 disclosing that it detected and blocked misuse of Claude models by state-linked and criminal actors. This included five case studies of attempts to use Claude in ways that could support biological weapons development (involving chikungunya, avian influenza, smallpox, and animal toxins), including a request from an actor affiliated with an unidentified military institution seeking help with a chikungunya research proposal that could have made the virus more dangerous. The report also identified Russia-linked cyber-espionage group GTG-20006 (linked to Midnight Blizzard), Chinese state-linked activity, and Yemen-based arms manufacturers using Claude for conventional weapons development, plus fraud, influence operations, and surveillance-tool cases. Anthropic said users in Russia, China, and Iran circumvented geographic restrictions via fraudulent accounts and resellers, and acknowledged that safeguards on older models (Haiku, Sonnet, Opus) were less stringent than on newer ones, cautioning it could not always determine whether flagged research was legitimate or nefarious.

X Corp sued New York over the 'Stop Hiding Hate Act,' which requires large social media companies to disclose how they monitor hate speech, extremism, harassment, foreign interference, and disinformation. On August 26, 2026, US District Judge John P. Cronan (SDNY) dismissed X's complaint with prejudice, ruling that requiring disclosure of 'purely factual and uncontroversial information' about content-moderation policy does not violate the First Amendment.

On August 24, 2026, French AI company Mistral AI announced a strategic collaboration with HUMAIN, the Saudi state-owned AI company, to jointly develop and localize frontier AI models optimized for Arabic, explore use of HUMAIN's Saudi data-center infrastructure, and pursue a joint go-to-market strategy across regulated sectors (financial services, telecommunications, public sector, manufacturing) in Saudi Arabia and the wider Middle East. The deal is valued in the 'hundreds of millions of euros' though key infrastructure and customer terms remain undisclosed. Mistral framed the deal around data 'sovereignty' and customer control, but made no public statement addressing human rights concerns about deploying AI infrastructure for a government with a documented record of surveillance and suppression of dissent, including for public-sector deployment.

Effective August 17, 2026, Atlassian's updated data contribution policy made metadata contribution (statistical characteristics, content patterns, search query keywords, Rovo Chat conversations) mandatory and non-optional for Free, Standard, and Premium tier customers; only Enterprise-tier customers retain the ability to opt out, and that ability is lost if an org downgrades from Enterprise. Atlassian's own support documentation states de-identified and aggregated data may be retained for up to seven years, and its sub-processor list names OpenAI as an authorized processor despite Atlassian's stated zero-data-retention agreements with third-party LLM partners.

On August 15, 2026, Bengaluru delivery workers organized by the Karnataka App-based Workers Union struck against Zomato (and Swiggy) in solidarity with a restaurant boycott threat led by the Bengaluru Hotel and Restaurant Association, which said over 21,000 restaurants could stop accepting orders. Workers and restaurants demanded an end to commissions as high as 30%, transparent payout and deduction systems, and dedicated human support to resolve disputes, calling platform economics exploitative of workers, restaurants, and customers alike.

On August 15, 2026, Bengaluru delivery workers organized by the Karnataka App-based Workers Union struck against Swiggy (and Zomato) in solidarity with a restaurant boycott threat led by the Bengaluru Hotel and Restaurant Association, which said over 21,000 restaurants could stop accepting orders. Workers and restaurants demanded an end to commissions as high as 30%, transparent payout and deduction systems, and dedicated human support to resolve disputes, calling platform economics exploitative of workers, restaurants, and customers alike.

Poland's Central Bureau for Combating Cybercrime launched an investigation on August 12, 2026 after MyDr, an electronic health records vendor used by over 10,000 Polish clinics, suffered unauthorized access exposing roughly 18.8 million patients' data (about half of Poland's population), including PESEL national ID numbers, names, contact details, diagnoses, and prescription records from 12,000+ facilities. Because MyDr is a data processor rather than a controller under GDPR, it could not directly notify affected patients, instead relying on client clinics to pass notifications along, delaying individual disclosure. MyDr said its systems were secure and that no evidence of public sale of the stolen data had been found as of the investigation's start.

Ola Electric reversed a Rs 57 crore (~$6.8M) provision it had booked for potential penalties under India's battery-cell Production-Linked Incentive scheme in its June-quarter (Q1 FY27) results, before receiving formal approval from the Ministry of Heavy Industries for the extension and waiver it had requested. Statutory auditor BSR & Co. LLP (a KPMG India affiliate) issued a qualified audit opinion - the company's first since its 2024 listing - stating it could not obtain sufficient appropriate audit evidence in the absence of MHI approval. Ola Electric shares fell 5-6% on the news despite a narrower quarterly loss.

India's SEBI issued a show-cause notice on August 11, 2026 to Paytm CEO Vijay Shekhar Sharma and CFO Madhur Deora concerning the timing of the company's December 6, 2023 disclosure that it would curtail small (sub-Rs 50,000) personal loans following RBI's tightened digital lending rules. Paytm shares fell roughly 12% in the eight trading days before the announcement and a further 20% the day after it was made, prompting scrutiny of whether the disclosure was delayed relative to when the decision was made. Paytm said it does not expect a financial impact from the notice and has 14 days to respond.

Hachette Book Group, Cengage Learning, Elsevier, and author Scott Turow filed a class action against Google LLC on July 10, 2026, alleging Google copied millions of copyrighted works to train its Gemini AI models. The complaint alleges Google misused books obtained through the Google Books program (provided only for search functionality), downloaded pirated works from shadow libraries and behind paywalls, and intentionally removed or altered copyright management information to conceal the source of the training data. An internal Google document cited in the complaint reportedly acknowledged the practice could be 'highly problematic' with potential exposure of '$10Bs-$100Bs in potential fines.'

Germany's financial regulator BaFin opened a formal review of Zalando's 2025 consolidated financial statements after finding concrete evidence the company breached accounting rules by omitting a required related-party transaction disclosure from the notes: that Anders Holch Povlsen, Zalando's largest shareholder, also held a significant stake in About You before Zalando's EUR1.13 billion ($1.29 billion) acquisition of it. Shares fell as much as 10.6% on the news before paring losses. Zalando called the omission 'purely formal and materially insignificant,' noting the underlying information was already public via the tender offer documents, and said it was in constructive dialogue with BaFin.

Ranking Digital Rights' 2026 Telco Giants Edition, published June 22, 2026, ranked MTN Group 2nd of the world's major telecom companies on digital rights disclosure (behind Telefónica), up from 6th place previously, making it the first emerging-market telco to reach the top 3. RDR credited MTN with strengthened governance disclosures, a new advertising content policy, and enhanced user-data-protection disclosures, while noting MTN still lacks detailed AI governance policies, including on AI risk review and whether user data is used to train AI models, and that its transparency gains face testing from the regulatory and political realities of the markets it operates in.

A proposed class action filed in the U.S. District Court for the Northern District of California on June 14, 2026 by D.C. resident Karl Kahn alleged that Anthropic's $200/month Claude Max 20x plan delivers only 6-8x Pro tier usage rather than the advertised 20x, and that the $100/month Max 5x plan delivers only 3.5x. The suit also challenges Anthropic's claim that Max 20x offers '50% savings'. Two days after the complaint, on June 16, 2026, Anthropic implemented metered credit caps for agent SDK, headless CLI, GitHub Actions, and third-party app usage, separating them from interactive subscription limits.

Tata Consultancy Services confirmed plans to cut approximately 12,200 jobs (~2% of its global workforce) during 2026, with managers instructed to identify the bottom 5% of staff as 'Band-D' underperformers - a process Indian IT-services labor advocates describe as a 'PIP-as-layoff playbook' designed to engineer voluntary exits at scale. The cuts disproportionately affect mid-career engineers in India and have triggered organizing among Indian IT services workers about their rights, MSA contracts, and background verification practices.

Senator Elizabeth Warren sent Nvidia a formal letter demanding it detail compliance measures after multiple DOJ enforcement actions alleged unlawful diversion of Nvidia H100/H200 chips (roughly $160M) and servers (roughly $510M) to China via transshipment through Malaysia and Thailand, and questioned CEO Jensen Huang's public statements that there was 'no evidence of any AI chip diversion' and that Nvidia's China market share had 'dropped to zero.' Warren later invoked statutory authority under the Export Control Reform Act to compel the Commerce Department to turn over related enforcement records, and jointly with Sen. Banks urged Commerce to suspend or reconsider Nvidia's export licenses following a related Supermicro diversion indictment.

Argentina's Buenos Aires provincial government opened a consumer-protection investigation in late May 2026 into MercadoLibre over allegedly abusive contract terms imposed on platform sellers and buyers, with regulators publicly indicating a fine of up to 1.8 billion pesos was under consideration. The probe followed earlier provincial enforcement against Rappi for undisclosed surcharges and is part of a broader Buenos Aires push to regulate platform contracts and undisclosed fees.

WiseTech announced roughly 2,000 layoffs in February 2026 as part of an AI-driven restructuring, with redundancy notifications beginning in May 2026 after an extended uncertainty period. The Guardian and Australian Financial Review reported that WiseTech sent staff in China a redundancy email titled 'Our global transformation - next steps', omitting any mention of AI, while staff elsewhere received an email titled 'Our AI Transformation - next steps'. CEO Zubin Appoo confirmed the differing communications, citing differing jurisdictional legal requirements (Chinese labor law makes AI-driven redundancies harder to justify than 'major changes' such as relocation or merger). A union-organized employee petition drew 590+ signatures citing prolonged uncertainty and minimum-only redundancy packages; a subsequent fact-check disputed claims that email body content (beyond the subject line) was altered.

A securities class action against Globant alleged the company failed to disclose wage freezes imposed on engineering staff in Argentina and Mexico in late 2023, resulting in widespread employee dissatisfaction, service-quality decline, and client losses that were not communicated to investors. The complaint claims investors suffered damages when the underlying Latin American workforce conditions became public.