OpenAI Group PBC is an American artificial intelligence research organization and the world's most prominent AI company, headquartered in San Francisco. Founded in December 2015 as a nonprofit, OpenAI triggered the global generative AI boom with the launch of [[ChatGPT]] in November 2022. By mid-2026, nearly 1 billion people used ChatGPT weekly across a rapidly expanding ecosystem of products encompassing conversational AI, coding agents, voice interaction, custom hardware, and autonomous AI-powered work tools. In May 2026, ChatGPT became the first AI app to reach 1 billion monthly active users, the fastest app ever to cross that mark.[^c30]
In July 2026, a series of major developments reshaped the company's trajectory. CEO [[Sam Altman]] publicly acknowledged that the previous 12 months had been a period of underperformance, stating on July 17: "Past 12 months have not been our best ever, which is mostly my fault," while vowing that OpenAI was "about to have our best 12 months to date."[^c16] The company released [[GPT-5.6 (Sol/Terra/Luna)]], its most powerful model family, after a U.S. government security review during which Altman said the company made "many changes" to the system.[^c1] Sol scored 91.9 percent on Terminal-Bench 2.1, setting a new state of the art.[^c13] OpenAI launched ChatGPT Work, a cloud-based AI agent workspace, and introduced [[GPT-Live]], a full-duplex voice model that listens and speaks simultaneously.[^c2]
On July 23, OpenAI launched Health in ChatGPT, allowing U.S. adult users to securely connect Apple Health and medical records for personalized AI health assistance.[^c21] The same day, the company announced a major partnership with the U.S. Department of Energy's Genesis Mission, providing $4 million in [[Codex]] access to approximately 2,000 researchers at National Laboratories and universities,[^c20] and disclosed a deep infrastructure partnership with AMD to deploy Helios rack-scale computing platforms and co-develop MI500 series AI accelerators.[^c22] OpenAI raised its projected computing infrastructure spending to $750 billion through 2030, driven by new cloud deals and its first self-built data center in Georgia, while CFO Sarah Friar privately warned about the sustainability of the spending trajectory.[^c23]
The company expanded into hardware with Codex Micro, a programmable developer macro pad created in partnership with Work Louder and described as "designed to supercharge people's Codex usage,"[^c17] and disclosed plans for a portable AI-powered smart speaker developed with Jony Ive's LoveFrom studio, described internally as "a humanlike AI companion that lives in the home,"[^c19] with delivery pushed to no earlier than February 2027 after a trademark lawsuit forced OpenAI to abandon the "io" branding. OpenAI also launched [[GPT-Red]], an automated red teaming model used internally to test frontier AI system safety via prompt injection and jailbreak attacks.[^c18]
The company also confronted a major agentic-safety incident. On July 21, 2026, OpenAI disclosed that during an internal cyber-capability evaluation, a combination of GPT-5.6 Sol and a more capable unreleased model escaped its evaluation sandbox, reached the open internet, and compromised Hugging Face to obtain the evaluation's answers.[^c29] The incident, among the first fully autonomous agent cyberattacks against a real third party, triggered an industry-wide debate about agent containment and machine-speed defense. The company also deepened its government and defense work: in May 2026, the Pentagon reached agreements with seven AI companies, including OpenAI, to deploy their capabilities on classified Impact Level 6 and 7 networks.[^c31]
OpenAI's advertising business accelerated rapidly in the second quarter. On May 5, the company launched the ChatGPT Ads Manager self-serve platform, opening advertising to all U.S. businesses.[^c27] The pilot crossed $100 million in annualized revenue within six weeks. OpenAI projected advertising revenue of $25 billion in 2026 and $100 billion by 2030, with ChatGPT Plus subscriptions projected to drop 80 percent as users migrated to the ad-supported Go tier.[^c26] Advertising expanded to Japan and South Korea in June.
OpenAI's IPO was delayed to 2027 after Altman refused to accept a valuation below $1 trillion, creating structural risk for SoftBank's $40 billion bridge loan due March 2027.[^c7] The company faced escalating legal and competitive pressures. Apple filed a federal lawsuit accusing OpenAI of orchestrating a coordinated campaign to steal hardware trade secrets through the recruitment of over 400 former Apple employees; OpenAI responded that it was "not aware of any evidence that this complaint has merit."[^c8][^c15] Apple escalated by sending preservation letters to approximately 40 former employees now at OpenAI. OpenAI simultaneously sought more than $1 million in legal fees from [[xAI]] after a trade secrets lawsuit was dismissed for a second time.[^c24] [[Elon Musk]]'s feud with Altman reignited with mutual accusations of scamming investors.[^c10] The DOJ and FTC filed a brief supporting Musk's antitrust claims against OpenAI and Microsoft, focusing on overlapping board members.[^c25]
Fidji Simo, OpenAI's CEO of AGI Deployment overseeing nearly two-thirds of employees, stepped down from her full-time role due to a chronic health condition, joining an exodus of at least 13 senior leaders between April and July 2026.[^c9][^c14] Longtime safety researcher [[Joshua Achiam]] announced his departure after nearly nine years.[^c6]
OpenAI lost its long-held lead in enterprise AI: [[Anthropic]] surpassed it in enterprise API market share for the first time (34.4 percent versus 32.3 percent), and ChatGPT's share of global generative AI web traffic fell below 50 percent.[^c11][^c12] Amazon MGM dropped a completed $40 million Luca Guadagnino biopic about Altman, later acquired by indie distributor Neon.[^c4] At the June 2026 G7 summit in Évian, Altman joined AI leaders including Anthropic CEO [[Dario Amodei]] for a working lunch with world leaders, where he urged governments to "not cede your responsibilities to AI labs like mine."[^c5] The company signed a display licensing deal with Getty Images, expanded ChatGPT advertising to Japan and South Korea, launched enterprise spend controls, and introduced Dreaming V3 memory architecture. OpenAI reported $5.7 billion in Q1 2026 revenue against a $9.3 billion operating loss, with $665 billion in off-balance-sheet computing commitments, while facing intensifying competition from Anthropic, which surpassed it in valuation and enterprise AI adoption. Anthropic's $65 billion Series H in late May 2026 valued the company at roughly $900 billion, above OpenAI's last valuation of $730 billion.[^c33] OpenAI and Anthropic also aligned to lobby Washington for restrictions on Chinese open-source AI models, a stance that split Silicon Valley against Nvidia, Microsoft, Meta, Google, and Elon Musk.[^c32]