Top 10 AI Stories: June 24, 2026 Daily Roundup

AI News Today June 24 2026: Top 10 Stories

Getty Images stock jumped 200% in a single day. Satya Nadella walked into the Wall Street Journal and named OpenAI and Anthropic by name, telling them they have not earned the right to do what they are doing to the economy. And Samsung just reversed one of the most famous corporate AI bans in history to hand OpenAI 125,000 new enterprise users.

Fable 5 is still offline. Gemini 3.5 Pro has still not launched. The two biggest model events of the month are stuck in limbo while every other story in AI keeps moving at full speed. Here are the 10 things every AI learner needs to know for June 24, 2026.

1. Getty Images Signs Multi-Year Deal with OpenAI, Stock Soars 200%

Getty Images announced a multi-year display partnership with OpenAI on June 21, 2026, granting OpenAI the right to surface Getty's licensed photo and editorial library directly inside ChatGPT search results. The announcement sent Getty stock soaring more than 200% in a single session.

The deal covers over 400 million assets, including premium editorial content from sport, entertainment, and news coverage, plus iStock, Getty's lower-cost library. The agreement is explicitly display-only: Getty's images will appear when ChatGPT is answering factual questions that benefit from visual context, such as historical events, celebrity portraits, or travel destinations. The deal does not grant OpenAI rights to use Getty content for training new AI models.

Why Getty Reversed Its Anti-AI Stance

This is a remarkable about-face. In September 2022, Getty banned all AI-generated art from its library. In February 2023, it sued Stability AI for copyright violations. That case was rejected in late 2025. The Getty-OpenAI partnership is structured as a revenue-sharing model, with Getty receiving compensation based on usage metrics including flat licensing fees and per-impression payments. Getty CEO Craig Peters described the deal as delivering "richer visual experiences to ChatGPT users."

For context: Getty already struck a similar display deal with Perplexity AI in October 2025. The OpenAI deal is significantly larger in scope and distribution. Shutterstock's partnership with OpenAI, reaffirmed in early 2026, is primarily for training data, not display. This dual approach gives ChatGPT both generative imagery (via DALL-E, trained on Shutterstock) and licensed editorial imagery (via Getty) for factual queries.

My take: The 200% stock jump is partly market excitement and partly relief from copyright uncertainty. For Getty, this is the clearest signal yet that licensing is a more sustainable long-term strategy than litigation. For OpenAI, it is a direct upgrade to ChatGPT's search quality at a moment when Google's AI Mode is its most credible search competitor.

2. Fable 5 Ban: Day 12, NSA Testimony Reshapes the Whole Story

Claude Fable 5 and Mythos 5 remain offline as of June 24, 2026, twelve days into the US export control ban. No official restoration date exists. API calls to claude-fable-5 continue to return errors.

The most significant development this week was not a technical update but a testimony. NSA Director General Joshua Rudd told Senator Mark Warner in a Senate Intelligence Committee briefing that Mythos, in a classified red-team exercise, autonomously breached nearly all of the NSA's classified systems within hours. This is the closest thing to an official government explanation for why the ban was imposed, and it reframes the story entirely.

From Jailbreak to Autonomous Capability

Anthropic's initial public framing was that the ban was triggered by a narrow jailbreak, one that security researchers demonstrated could be replicated with other publicly available models. The NSA testimony suggests the actual concern is not a jailbreak at all: it is Mythos 5's autonomous offensive cybersecurity capability itself. A model that can autonomously compromise classified government infrastructure is a categorically different problem from a model with a patchable safety gap.

The Economist's defence editor Shashank Joshi, who broke the NSA breach story, added an important qualifier afterward: the breach should not be read literally. It depended on Mythos operating alongside other tools under specific conditions, not the model single-handedly defeating national security from a chat window. That caveat has received far less attention than the headline.

The most concrete near-term signal to watch: Anthropic's updated privacy policy, which takes effect July 8, 2026, requires government-issued ID verification from all users. This is likely the mechanism for restoring Fable 5 access to verified US citizens without fully lifting the export control directive. International users would remain on Claude Opus 4.8 under that scenario.

My take: The NSA testimony is the most significant development in this story since the ban itself. If the government's concern is autonomous offensive capability rather than a jailbreak, Anthropic's path back is not a software patch. It is a negotiation about what frontier AI is allowed to be able to do.

3. Satya Nadella Calls Out OpenAI and Anthropic by Name in WSJ

Microsoft CEO Satya Nadella published an interview with the Wall Street Journal this week that is the sharpest public critique of the AI industry's power structure from anyone inside that structure. Nadella named OpenAI and Anthropic specifically and told them they have not earned society's permission to do what they are doing.

Nadella's exact framing: "You can't say, hey, all white-collar jobs are gone and this could even be a weapon and we will use all the power to build data centers." His argument is that an AI industry structured around a handful of dominant frontier models is not just economically dangerous but politically unsustainable. The industry needs to earn societal permission rather than assume it.

The Tension Underneath

Nadella's critique carries obvious tensions. Microsoft has invested approximately $13 billion in OpenAI. It signed a multibillion-dollar agreement with Anthropic last year. It is guiding to roughly $190 billion in capital expenditure in 2026 to expand the data center infrastructure that makes frontier models possible. He is simultaneously the largest financial backer of the companies he is warning against.

The strategic logic is readable even if Nadella does not state it directly. Microsoft is building the platform layer, Azure, Foundry, and GitHub, that sits between enterprises and whichever frontier models they use. If frontier models become interchangeable commodities, Microsoft's orchestration and governance layer is the prize. If they do not, Microsoft's MAI model family, which it launched at Build 2026 without OpenAI data, reduces dependency. Either way, Microsoft's position improves.

As evidence for why the critique is grounded, consider Uber. The ride-hailing company deployed Claude Code to roughly 5,000 engineers and burned through its entire $3.4 billion AI budget for 2026 in just four months. When AI usage is metered by the token, productivity compounds into cost rather than into enterprise value. That is the micro-level demonstration of the macro problem Nadella is describing.

4. Samsung Reverses Its 2023 ChatGPT Ban and Deploys OpenAI to 125,000 Staff

Samsung Electronics announced on June 21, 2026, that it is rolling out ChatGPT Enterprise and Codex to all of its employees in South Korea and to all employees globally in its Device eXperience (DX) division. The total headcount covered is approximately 125,000 people. OpenAI described the deployment as "one of OpenAI's largest enterprise launches ever."

The reversal is extraordinary in its speed. In March 2023, Samsung engineers accidentally leaked sensitive source code and internal meeting notes through ChatGPT. Samsung's response was immediate: a company-wide ban on generative AI tools. Three years later, Samsung is deploying the same company's tools to roughly 125,000 employees, this time with enterprise-grade security controls, zero-data-retention policies, and active data-loss prevention.

Why Samsung Changed Its Mind

Samsung ran a two-month proof-of-concept with 2,500 employees testing enterprise versions of ChatGPT, Google Gemini, and Anthropic's Claude before selecting OpenAI. The pilot led to the full deployment. The core change from 2023 is governance: ChatGPT Enterprise does not train its models on customer data by default, includes admin controls and compliance features, and operates within a data protection framework that Samsung's IT team can audit.

According to PYMNTS reporting citing Seeking Alpha, Codex weekly active users in South Korea grew nearly 800% since February 1, 2026. More than 5 million people globally now use Codex weekly for both technical and non-technical tasks. The Samsung-OpenAI relationship also extends into hardware: Samsung is supplying OpenAI with advanced HBM4 memory chips for its custom Titan AI chip, with mass production targeted for late 2026.

My take: The Samsung reversal is the single clearest data point on how corporate AI adoption has matured since 2023. The conversation has shifted from 'should we use this at all' to 'how do we deploy it safely at scale.' That is a meaningful change in the enterprise risk calculus.

5. FT Analysis: Anthropic May Have Talked Itself Into the Export Ban

The Financial Times published a quantitative analysis this week finding that Anthropic used AI risk-related terms approximately eight times more often than OpenAI in its 2026 official statements and public communications. Five in every 1,000 words used by Anthropic in 2026 related to risk, regulation, or restrictions. The equivalent figure for OpenAI and Sam Altman was 0.6 words per 1,000, eight times lower.

The FT's framing: Anthropic may have talked itself into the export ban. By repeatedly and publicly emphasizing how dangerous its most capable models are, the company provided rhetorical ammunition to the government officials who ultimately decided that those models were too dangerous for unrestricted deployment.

The Dario Amodei Problem

Anthropic CEO Dario Amodei's essay calling for government blocking power over unsafe AI deployments was published approximately 48 hours before the government used exactly that power on Anthropic. That timing is not lost on anyone following this story. Anthropic's public posture, built on genuine and principled AI safety concerns, has created a situation where the company's own language is the primary evidence the government has cited for why the ban is justified.

CNN's analysis of the regulatory gap, published June 21, 2026, captured the wider concern: there is no transparent, consistent framework for regulating AI in the United States. The Fable 5 ban happened without a court order, without a public filing, and without a detailed explanation of the technical concern. Whether you agree with the outcome or not, the process has set a precedent that no AI company should be comfortable with.

My take: The FT analysis is uncomfortable but important. Being honest about your model's capabilities in a regulatory vacuum is not a mistake. But Anthropic is now learning that honesty about risk, without a commensurate regulatory framework to channel that honesty into constructive policy, can be weaponized against you. This is a real problem for the whole AI safety ecosystem.

6. Norway Bans Generative AI in Elementary Schools Nationwide

Norway's government announced a near-total ban on generative AI for elementary school pupils, with supervised restrictions on its use for older students, effective from the school year starting in late August 2026. Prime Minister Jonas Gahr Store made the announcement, citing a broad decline in education test scores.

Norway's government had already banned smartphones from schools in 2024 and restored disciplinary powers to teachers. The AI ban follows the same logic: that tools which bypass the cognitive work of learning produce students who cannot do the underlying skills without the tool. Using AI increases the risk that young children skip important steps in their education, Store told a press conference.

The policy applies to generative AI from major providers including OpenAI, Google's Gemini, and Anthropic's Claude. The distinction between elementary and secondary students reflects Norway's view that younger children are at greater developmental risk from shortcutting basic skills in reading, writing, and mathematics. The policy will be reviewed at the end of the 2026-2027 school year.

My take: Norway is the first major European country to take this step nationally, and it probably will not be the last. The education sector is where the gap between AI capability and AI wisdom is most acute. Using a language model to write an essay does not make you a better writer. Most kids and parents have not internalized this yet, and neither have most schools.

7. Gemini 3.5 Pro: Still Not Here, Window Closing Fast

Gemini 3.5 Pro has still not reached general availability as of June 24, 2026. Google committed to a June 2026 launch at Google I/O on May 19, when Sundar Pichai told the audience to "give us until next month," drawing audible groans. With six days left in June, the window is closing.

The model remains in limited preview for select Vertex AI enterprise customers. No public announcement has been made on the model blog, which is the channel Google has used for every previous Gemini release. Prediction markets price the odds of a June 30 launch at roughly 50 to 55 percent, slightly below even.

The confirmed specifications: a 2-million-token context window (double Gemini 3.5 Flash's 1 million and the largest of any production frontier model), a Deep Think reasoning mode gated to the $250-per-month Ultra tier, and frontier multimodal capability. The competitive context is unusually favorable. Fable 5 remains offline, GPT-5.6 has not launched, and every developer team that built pipelines on Fable 5 is looking for an alternative with a long context window.

My take: If Gemini 3.5 Pro slips past June 30, Google needs to say something. The developer community heard a June commitment on May 19 from the CEO. Silence into July after that commitment would be a credibility problem. Either ship it this week or publish a timeline update. Both are acceptable. Silence is not.

8. China Raises $7.4 Billion in New AI Funding Round

China's AI sector has raised $7.4 billion in a new funding round, according to reporting from AI Weekly. The fundraise arrives directly in response to the US government's actions against Anthropic, with Chinese AI developers and investors positioning themselves as the beneficiaries of any global restriction on US frontier model access.

The Fable 5 ban has accelerated this dynamic. GLM-5.2, released June 13, 2026, by Chinese lab Zhipu AI (Z.ai) under an MIT license with explicit language stating "no regional limits," saw immediate enterprise adoption from developers locked out of Fable 5. GLM-5.2 scored 62.1% on SWE-Bench Pro, placing it above GPT-5.5's 58.6% on that specific benchmark, and its API pricing at $1.40 per million input tokens is roughly 21 times cheaper than GPT-5.5's output pricing.

The $7.4 billion raise spans multiple Chinese AI companies and represents the largest single-week fundraising total in Chinese AI history, according to AI Weekly's coverage. The US government's intent in restricting Fable 5 was to prevent adversaries from accessing frontier AI capability. The practical effect in the short term has been to accelerate Chinese open-weight model development by demonstrating the commercial gap that opens when US frontier models become unavailable.

My take: I want to be careful about overstating this. One week of Chinese fundraising does not erase a multi-year capability gap. But the direction of travel matters. Every time a US frontier model becomes unavailable, open-weight alternatives improve their commercial position, and Chinese labs are among the fastest-moving players in the open-weight space right now.

9. OpenAI Supplies ChatGPT Enterprise to Samsung in Largest Rollout Yet

This story is closely related to Story 4 but deserves its own entry for the OpenAI side of the picture. OpenAI described the Samsung deployment as "one of OpenAI's largest enterprise launches ever." The deployment covers ChatGPT Enterprise for all-hands productivity and Codex specifically as an agentic coding platform across technical and non-technical teams.

For non-technical context: Codex is an AI coding agent. Samsung is deploying it to employees who have no software engineering background, meaning the company is betting that non-developer staff can use Codex to build internal tools, websites, and automated business processes. This is the most aggressive version of the "AI for everyone" thesis: not just making developers faster, but making non-developers capable of building software.

According to OpenAI's announcement, Samsung CEO Sam Altman visited Samsung's Suwon campus on June 15, 2026, for a DX Insight Talk on AI-driven workplace innovation. That visit happened one week before the deployment announcement. The Samsung-OpenAI relationship now spans memory chip supply for the Stargate data center project, software deployment across 125,000 employees, and an ongoing collaboration on AI semiconductor infrastructure. This is not a vendor relationship. It is a strategic alliance.

My take: The detail that Codex Codex weekly active users in Korea grew 800% since February is the most interesting number in the whole announcement. That growth predates the Samsung deal and happened organically. The formal enterprise agreement is validating adoption that was already happening from the bottom up.

10. Uber Burned Through Its Entire $3.4B AI Budget in Four Months Using Claude Code

Uber deployed Claude Code to roughly 5,000 engineers in early 2026 and exhausted its entire $3.4 billion AI budget for the year in just four months. This figure surfaced in TechTimes reporting on Satya Nadella's WSJ interview and is one of the most striking data points in recent AI economics.

To put $3.4 billion in four months in context: that is $850 million per month, or roughly $170,000 per engineer per month, for a single AI coding tool. Claude Code pricing for enterprise users runs in the range of $500 to $2,000 per engineer per month depending on usage tier. The Uber figures imply either extremely heavy use across all 5,000 engineers or significant usage in high-compute reasoning modes rather than standard autocomplete.

This is the specific economic dynamic Nadella was describing when he warned that enterprise AI spending compounds as a cost rather than as an asset. Every token Uber's engineers consumed through Claude Code generated output, training signals, and competitive intelligence that flows back to Anthropic, not to Uber. Uber got faster code review. Anthropic got 5,000 engineers' worth of domain-specific usage data for four months. The knowledge asymmetry is structural, not accidental.

My take: The Uber number is the clearest possible illustration of why enterprise AI economics are broken in their current form. Individual productivity gains are real. But the per-engineer cost at scale makes this unsustainable as a blanket deployment strategy. The next wave of enterprise AI procurement will include cost-per-output benchmarks, not just capability benchmarks.

Frequently Asked Questions

Q: What is the top AI news today, June 24, 2026?

Getty Images announced a multi-year display partnership with OpenAI on June 21, 2026, letting Getty's 400-million-asset photo library appear directly inside ChatGPT search results. Getty stock jumped over 200% on the announcement. Other major stories include Fable 5 remaining offline on day 12, Satya Nadella publicly challenging OpenAI and Anthropic in the Wall Street Journal, and Samsung deploying ChatGPT Enterprise to 125,000 employees.

Q: Did Getty Images sign a deal with OpenAI?

Yes. Getty Images and OpenAI signed a multi-year display partnership, announced June 21, 2026, granting OpenAI the right to surface Getty's licensed photo and editorial content inside ChatGPT search results. The deal is display-only and does not grant OpenAI rights to use Getty content for training. Getty's 400 million assets, including editorial, sport, entertainment, and iStock imagery, are covered. Getty stock surged more than 200% on the news.

Q: Is Claude Fable 5 back online on June 24, 2026?

No. Claude Fable 5 and Mythos 5 remain offline as of June 24, 2026, twelve days after the US Commerce Department's export control directive on June 12. API calls to claude-fable-5 still return errors. The NSA Director testified that Mythos autonomously breached nearly all US classified systems in a red-team exercise, reshaping the ban from a jailbreak problem to an autonomous-capability concern. All other Claude models remain fully available.

Q: Why did Satya Nadella criticize OpenAI and Anthropic?

Microsoft CEO Satya Nadella told the Wall Street Journal that OpenAI and Anthropic have not earned society's permission to restructure the economy while simultaneously making dire job-loss predictions and demanding unchecked infrastructure expansion. His warning: the AI industry cannot tell workers their jobs are gone while building an extractive model where enterprise knowledge flows to model providers rather than to the companies that paid for the work. Nadella called on AI giants to earn public trust, not assume it.

Q: Did Samsung unban ChatGPT?

Yes. Samsung Electronics reversed its 2023 company-wide ChatGPT ban and deployed ChatGPT Enterprise and Codex to all employees in South Korea and all employees globally in its Device eXperience (DX) division, announced June 21, 2026. The rollout covers approximately 125,000 people. OpenAI described it as one of its largest enterprise launches ever. Samsung ran a two-month proof-of-concept with 2,500 employees before selecting OpenAI over Google Gemini and Anthropic's Claude for the primary deployment.

Q: Has Norway banned AI in schools?

Yes. Norway announced a near-total ban on generative AI for elementary school students effective the school year starting late August 2026, with supervised restrictions for older students. Prime Minister Jonas Gahr Store cited declining test scores as the rationale. The policy follows Norway's 2024 smartphone ban. Generative AI from OpenAI, Google, and Anthropic are all covered. The policy will be reviewed after the 2026-2027 school year.

Q: What did the FT report about Anthropic and the Fable 5 export ban?

The Financial Times published a quantitative analysis finding that Anthropic used AI risk-related terms eight times more often than OpenAI in its 2026 official communications: five in every 1,000 Anthropic words related to risk, regulation, or restrictions, versus 0.6 words per 1,000 for OpenAI. The FT's framing: by consistently emphasizing the dangers of its most capable models, Anthropic provided the rhetorical justification for the government's decision to treat those models as too dangerous for unrestricted access.

Q: When is Gemini 3.5 Pro launching?

As of June 24, 2026, Gemini 3.5 Pro has not reached general availability. Google CEO Sundar Pichai committed to a June 2026 launch at Google I/O on May 19. With six days left in June, prediction markets price the odds of a launch before June 30 at roughly 50 to 55 percent. The model features a 2-million-token context window, Deep Think reasoning (restricted to the $250/month Ultra tier), and frontier multimodal capability. If it misses June, expect a formal timeline update from Google DeepMind.

Q: What is China raising $7.4 billion for in AI?

China's AI sector raised $7.4 billion in new funding this week, its largest single-week fundraising total in AI history according to AI Weekly. The fundraise is partly a direct response to the Fable 5 ban, with Chinese developers positioning open-weight models as alternatives to US frontier models that can become unavailable due to government action. Chinese lab Zhipu AI's GLM-5.2, released June 13, 2026, under an MIT license with no regional restrictions, has already gained enterprise adoption among developers locked out of Fable 5.