September 2026
The month's biggest shifts in AI and video generation, the money behind them, and the people and labs in the news.
A price war at the top
Anthropic's Claude Opus 5.5 and OpenAI's GPT-6 Sol and Luna shipped on the same day, each far cheaper than the model before.
Nvidia buys Hugging Face
The open hub for AI models goes to the chipmaker. The deal should close in the first half of 2027.
Sora is gone
OpenAI switched off the Sora video API for good on 24 September.
Models & labsLabel your AI actors
A new California law says ads using AI-generated people or voices must say so.
Video generationAI video moves into familiar tools
Google put its video model into Vids and YouTube. Runway now works inside Premiere, After Effects and DaVinci Resolve.
Creator toolsAI video is real money
Higgsfield is on a $1B+ yearly pace, Synthesia is expanding in the US, and Katzenberg is opening an AI animation studio.
Nvidia buys Hugging Face for $12.9B
The company that makes most of the world's AI chips agreed to buy the open hub where developers share AI models. Nvidia promises to keep Hugging Face open, and the whole team is joining.
- purchase price
- $12.9B
- models hosted
- 3M+
- developers
- 18M+
- expected to close
- H1 2027
Why it's the story of the monthAlmost every open AI model, from video to voice, is shared through Hugging Face. Who owns it now shapes who gets access to open AI, and on what terms.
Where the industry gathered
The conferences, keynotes and festivals where people met in person this month.
2–12 Sept83rd Venice International Film Festival
One of the first AI-generated feature documentaries premiered at a major film festival, which set off a public argument over AI in cinema.
- AI-built Renzo Rosso doc 'Be Brave' premiered 10 Sept
- George Clooney spoke out against AI at the festival
- 'Woman Unknown' by May el-Toukhy won the Golden Lion
9 SeptApple Event: Surprise and Shine
Apple put Siri AI and new camera hardware into the phones that most creators shoot on.
- John Ternus's first September keynote as Apple CEO
- Foldable iPhone Duo launched from $1,999
- iPhone 18 Pro adds a mechanical variable-aperture 48MP camera
10–20 SeptToronto International Film Festival
Audiences openly pushed back on AI in filmmaking, and the festival launched its first official film market.
- Danny Boyle jeered at 'Ink' Q&A over AI-animated photos
- 'AI slop' stickers appeared on the film's billboard
- First official TIFF Market drew 1,473 buyers, up 19%
IBC2026
The world's biggest broadcast and video-tech show focused on using AI in real production work and proving where content comes from.
- 1,300+ exhibitors, 256 of them first-timers (up 36%)
- Accelerator projects covered agentic AI and content provenance
- Netflix, YouTube, Google and Meta spoke on AI in practice
17–20 SeptTokyo Game Show 2026
Generative AI took centre stage at Asia's biggest games show, and a new survey showed most Japanese game developers already use it.
- 85.8% of Japanese game developers now use generative AI
- An AI technology pavilion replaced the indie showcase
- Record 1,138 exhibitors; the last day was cancelled for a typhoon
23 SeptMade on YouTube 2026
YouTube's yearly creator showcase brought AI editing, live dubbing and likeness protection to millions of creators.
- Chat-based AI editing arrives in Shorts and YouTube Create
- Real-time AI auto-dubbing for livestreams
- Video A/B tests of up to three cuts, plus voice likeness detection
23–24 SeptMeta Connect 2026
Meta built its whole hardware lineup around Muse, its AI agent, and added camera glasses aimed at first-person video.
- Zuckerberg: Muse will become 'personal superintelligence' for billions
- Meta VR Glasses (~100g, $1,299.99) ship spring 2027
- Ray-Ban Meta Gen 3 records 3K video
Models & labs
New models, the labs behind them, and the rules they answer to.
Anthropic launches Claude Opus 5.5
Opus 5.5 performs about as well as Claude Fable 5.1 on most work and costs about 40% less to run than Opus 5, at $4/$20 per million tokens. Sonnet 5.5 and Haiku 5.5 are due in the coming weeks.
Why it mattersTop-tier AI writing and reasoning just got about 40% cheaper, so apps built on it can do more for the same price.
Anthropic ↗OpenAI releases GPT-6 Sol and GPT-6 Luna
Two cheaper models bring GPT-6 Astra's gains to lower price tiers. Sol costs $2/$10 per million tokens, half its old price, and Luna costs $0.10/$0.50. OpenAI says Sol makes about half as many factual mistakes as the model it replaces.
Why it mattersLuna is cheap enough that everyday tasks like summarizing, tagging and captioning cost almost nothing.
OpenAI ↗
OpenAI rolls out GPT-6 Astra with extra safeguards
OpenAI began rolling out its flagship GPT-6 Astra with additional safety measures added after the Hugging Face breach. The CFO also told staff that business customers now bring in more revenue than consumers.
Why it mattersOpenAI's flagship now powers the most demanding agent apps, and its business customers now outspend consumers.
CNBC ↗
Google ships Gemini 3.8 Flash TTS and Flash-Lite TTS
Two new text-to-speech models offer 2,000+ voices in 100+ languages, can design a voice from a written description, and can clone a voice from a 30-second sample. You can direct delivery line by line, including whispers, laughs and sighs.
Why it mattersRealistic, directable voices in 100+ languages are now a cheap building block for narration, dubbing and audiobooks.
Google ↗
California requires disclosure when ads use AI-generated performers
Governor Newsom signed SB 1050 at SAG-AFTRA headquarters. It requires a clear disclosure on any video or audio ad that uses a realistic AI-generated person or voice. Ads that break the rule can be pulled.
Why it mattersAnyone making ads with AI people or voices for California will need a clear label from next year.
Office of the Governor of California ↗Video generation
The models and tools that make video.
Sora API shuts down for good
OpenAI's Sora video API was switched off on Sept 24, five months after the Sora app closed. OpenAI announced the shutdown in March, after reports that Sora cost about $1M a day to run.
Why it mattersA major video model is gone, so apps built on it had to switch. Relying on a single AI provider is a real risk.
OpenAI Help Center ↗
Gemini Omni 1.1 Flash comes to Google Vids
Google Vids now uses Gemini Omni 1.1 Flash, which adds scene extension with smooth transitions, 1080p generation and upscaling, and exact control over clip length. It is rolling out across Workspace plans and to AI Pro/Ultra personal accounts.
Why it mattersMillions of office users now get story-style AI video inside a tool they already pay for.
Google Workspace Updates ↗
Google launches Gemini 3.8 Live with Live Avatar
A near real-time talking avatar with lip-sync and natural expressions, generated from a single image, is now generally available to Gemini Enterprise customers. It supports 97 languages and can hold a conversation for a few minutes at a time.
Why it mattersReal-time talking avatars now come from Google itself, putting pressure on specialists like HeyGen and Synthesia.
Google ↗
Runway previews GWM Worlds 2, a real-time interactive world model
Runway's research preview generates explorable worlds live at 720p and 24 fps, with 48kHz audio, which you steer with text actions and camera movement. There is no public access or pricing yet.
Why it mattersVideo may be turning into something you steer live, like a game, instead of a clip you render and watch.
Runway ↗Runway moves into Premiere, After Effects and DaVinci Resolve
Runway released plugins for Adobe Premiere Pro and After Effects on Sept 8 and a DaVinci Resolve plugin on Sept 23. It also added Kling and MiniMax H3 Max models to its platform, turning Runway into a hub for many different models.
Why it mattersAI video is moving into the editors professionals already use, instead of asking them to switch apps.
Releasebot (Runway release notes) ↗OpenAI releases ChatGPT Images 2.5 (GPT Image 2.5)
The new image model generates up to 50% faster, keeps reference subjects more consistent across edits, and adds Sketch-to-image. The API offers it in two versions: fast 'Flare' and detailed 'Sunburst'. OpenAI says users generate more than 3 billion images a week.
Why it mattersCharacters now stay more consistent from image to image, which has long been the hardest part of AI storytelling.
OpenAI ↗Creator tools
The apps and platforms people make video with.
Made on YouTube: AI editing assistant and episodic Shorts Series
YouTube added a chat-based editing assistant powered by Gemini Omni to Shorts and the Create app. It can trim pauses, reorder clips for pacing and add on-screen text. Creators can also group Shorts into seasons and episodes, aimed at the fast-growing microdrama format.
Why it mattersThe biggest video platform now edits for creators for free and supports series of episodes, a new home for serialized stories.
Google ↗Higgsfield says it is running at more than $1B a year
The AI video platform, which bundles many models like Veo, Kling and Seedance under one subscription, says its latest four weeks of revenue work out to over $1B a year. Contracted revenue from business customers grew 10x since June, driven by direct-to-consumer ad makers.
Why it mattersAn app built on top of other companies' video models has reached huge scale, and advertisers are the ones paying.
Bloomberg ↗
Synthesia opens a 50,000 sq ft US headquarters in New York
About 50% of Synthesia's revenue now comes from the US, and 90% of the Fortune 100 are customers, with 140% net revenue retention. US customers have moved from using it mainly for training videos to company-wide communication.
Why it mattersAI avatar video has become everyday business software, used by 90% of the Fortune 100.
Unite.AI ↗
HeyGen adds Edit Look and a real-estate product
HeyGen's monthly update, published Sept 9, lets users retouch and restyle an avatar without filming again, and adds a product for real-estate agents. HeyGen is reported to be at about $200M ARR.
Why it mattersAvatars can now be restyled after they are made, and AI video is being packaged for specific jobs like real estate.
HeyGen ↗
Utopai Studios and DeNA team up on AI-assisted anime
DeNA will use Utopai's storytelling AI system, PAI, in its animation workflows in Japan for visual development, character design and scene planning. Artists keep final approval on every creative decision.
Why it mattersJapan's anime industry is adopting AI on the terms that AI assists and the artist decides.
Variety (via Yahoo) ↗Deals & funding
Who raised money, and who bought whom.
Funding rounds 8
Investors put money into a company in exchange for a share of it.
DeepSeek raises new funding
- Raised
- ~$7.5B
DeepSeek now valued at ~$75B
The Chinese open-weights lab is reportedly finalizing a second outside round, expected to close by end of October, as its annualized revenue hit about $1B.
Samsung Electronics backs Mistral AI
- Raised
- €3B
Mistral AI now valued at €21B+
The French AI lab raised the largest equity round ever by a European tech company, largely to build and own data centers.
Scaleup Europe Fund (EQT)
PSG Equity
BlackRock
Andreessen HorowitzASML
.png)
General Catalyst
LightspeedNvidia
Salesforce Ventures+0 more
Andreessen Horowitz backs Cognition
- Raised
- $2B
Cognition now valued at $48B
The maker of the Devin AI coding agent nearly doubled its valuation in four months as run-rate revenue neared $900M.
Accel
Founders FundGeneral CatalystAvenir+0 more
Diffusion backs Harvey
- Raised
- $550M
Harvey now valued at $15.5B
The legal AI company, now past $400M in annual recurring revenue, raised again months after a round at $11B.
Lightspeed Venture Partners
Sequoia
Kleiner PerkinsAndreessen Horowitz
CoatueGGIC
WndrCo+0 more
Lightspeed Venture Partners backs Temporal
- Raised
- $550M
Temporal now valued at $12.55B
Temporal makes software that keeps long-running jobs (like AI agents) running reliably through failures; its valuation rose 2.5x in seven months.
Wellington Management
Goldman Sachs Alternatives
Andreessen HorowitzSequoia Capital+0 more
Insight Partners backs Snorkel AI
- Raised
- $350M
Snorkel AI now valued at $3.5B
Snorkel sells ready-made training datasets to AI labs; its annualized revenue grew 18-fold in a year to $375M.
S32
AdditionLightspeedGreylock
GV
Wells Fargo+0 more
Multiples Alternate Asset Management backs Brahma AI
- Raised
- $150M
Brahma AI now valued at $2B
Brahma, majority-owned by VFX house DNEG, is an AI operating system for enterprise video and audio used by Warner Bros and the NBA; the round may reach about $200M.
· Quartz ↗
Y Combinator backs VideoGen
- Raised
- $3.3M
VideoGen valuation not disclosed
An AI video platform whose agents turn a brief or uploaded media into editable, copyright-free videos with AI voiceovers; it has passed 5 million users.
Rebel Fund
Lobster CapitalSEStretford End Capital
Mento VC
Pioneer FundDecacorn VC+0 more
Acquisitions 3
One company buys another outright.
Nvidia buys Hugging Face
- Price
- $12.9B
Hugging Face is now part of Nvidia
Nvidia agreed to buy Hugging Face, the open hub where developers share AI models and datasets, promising to keep it an open platform; closing expected in H1 2027 pending regulators.
OpenAI buys Glass Imaging
- Price
- $300M+
Glass Imaging is now part of OpenAI
OpenAI reportedly bought Glass Imaging, a startup founded by ex-Apple Portrait Mode engineers that uses AI to improve smartphone photos as they are taken.
Lightstorm Vision buys Outsyders
- Price
- Undisclosed
Outsyders is now part of Lightstorm Vision
James Cameron's Lightstorm Vision bought Outsyders, whose AI-driven pipeline converts film into stereoscopic 3D (Lilo & Stitch, Tron: Ares); price undisclosed.
Partnerships 2
Two companies sign a large commercial agreement.
Anthropic and Akamai sign a deal
- Value
- $11.6B
Multi-year agreement
Anthropic will pay Akamai $11.6B over seven years for distributed cloud computing (expandable by $9B) and gets warrants for up to ~5% of Akamai; the largest deal in Akamai's history.
· Akamai ↗
ElevenLabs and Universal Music Group sign a deal
- Value
- Undisclosed
Multi-year agreement
UMG and ElevenLabs signed a multi-year licensing deal (ElevenLabs' first with a major label) to build a licensed AI platform where fans remix and mash up participating artists' tracks; terms undisclosed.
People in the news
The people behind this month's biggest moves.
James Cameron
Filmmaker, Lightstorm Vision
Cameron's Lightstorm Vision acquired Outsyders, an AI-driven 3D conversion studio, to scale 3D content production.
· Deadline ↗- Company
- Lightstorm Entertainment (co-founded 1990)
- Sept 2026
- Lightstorm Vision bought Outsyders
Andrew Tulloch
Researcher, Anthropic
The star researcher Meta reportedly courted with a package up to $1.5B left Meta's superintelligence TBD Lab after about 11 months to join Anthropic.
· Forbes Australia ↗- Now
- Anthropic, inference & performance team
- Previously
- Meta TBD Lab; co-founder, Thinking Machines Lab; OpenAI; 11 years at Meta
Arthur Mensch
Co-founder & CEO, Mistral AI
Mensch closed Europe's largest-ever tech equity round and said Mistral will pass $1B in annual recurring revenue before year-end.
· CNBC ↗- Co-founders
- Guillaume Lample, Timothée Lacroix
- Sept 2026
- Raised €3B Series D
Clément Delangue
Co-founder & CEO, Hugging Face
Delangue agreed to sell Hugging Face to Nvidia for $12.9B, saying his company approached Jensen Huang weeks earlier and the whole team will join Nvidia.
· CNBC ↗- Co-founded
- Hugging Face (2016)
- Co-founders
- Julien Chaumond, Thomas Wolf
- Sept 2026
- Agreed sale to Nvidia for $12.93B
Dario Amodei
Co-founder & CEO, Anthropic
Amodei and Anthropic's six other co-founders are seeking super-voting shares giving them 50.1% combined voting control ahead of an IPO.
· TechCrunch ↗- Co-founded
- Anthropic (2021)
- Sept 2026
- Founders seek 50.1% voting control before IPO
Jeff Dean
Co-founder, Discovery Loop
Weeks after leaving Google, Dean's new AI-for-science startup was reported to be in talks to raise at about $50B, five times its first valuation.
· Tech Funding News ↗- Google chief scientist
- 2023–2026
- Now
- Co-founder, Discovery Loop
- Known for
- MapReduce, Bigtable, Spanner, TensorFlow
Jeffrey Katzenberg
Co-founder, DreamWorks Animation & WndrCo
After a Sept 23 essay urging Hollywood to embrace AI (with credit, consent and compensation for artists), Katzenberg said he is launching an AI animation studio to make film and TV; no name, funding or date yet.
· TheWrap ↗- Born
- 1950
- Co-founded
- DreamWorks Animation, WndrCo
- Earlier
- Chairman of Walt Disney Studios, 1984–1994
- Also founded
- Quibi
Karandeep Anand
Chief Technology Officer, Disney
Disney named the former Character.AI CEO its first-ever CTO, overseeing technology, data and AI platforms and reporting to CEO Josh D'Amaro.
· The Walt Disney Company ↗- Previously
- CEO, Character.AI; 15 years at Microsoft
- Reports to
- Josh D'Amaro
Sam Altman
Co-founder & CEO, OpenAI
Altman said OpenAI won't go public in 2026, calling it an 'ill-advised' moment and citing a focus on safety.
· TechCrunch ↗- Co-founded
- OpenAI (2015), Loopt
- Sept 2026
- Ruled out a 2026 IPO
Labs in the news
What each AI lab did this month, in one place.
OpenAI
The company behind ChatGPT and the GPT models.
- Founded
- 2015
- CEO
- Sam Altman
- Latest valuation
- $852B post-money (March 2026); reportedly seeking up to $1.5T
- Latest round
- $122B committed (March 2026)
Anthropic
The AI lab behind the Claude models, now the most valuable AI startup.
- Founded
- 2021
- CEO
- Dario Amodei
- Latest valuation
- $965B post-money (Series H, May 2026)
- Total raised
- ~$125B
Mistral AI
France's leading AI lab, pitching 'sovereign' AI to European governments and firms.
- Founded
- 2023
- CEO
- Arthur Mensch
- Latest valuation
- €21B+ (Series D, Sept 2026)
- Latest round
- €3B Series D
DeepSeek
A Hangzhou lab making low-cost open-weight language models.
- Founded
- 2023
- CEO
- Liang Wenfeng
- Owner
- High-Flyer (hedge fund)
- Annualized revenue
- ~$1B (Sept 2026)
More companies in the news
Company news this month that the sections above didn't cover.
Suno launches v6 music models with Warner, BMG and Believe
Suno released v6 in three versions: v6, v6-wild and v6-mini, and v6-mini is the faster model on the free tier. New features include editing one section of a song without changing the rest, mashups from several sources, and prompts that can be text, audio, images or video.
New Pika puts video, image and sound in one app with 25+ tools
Pika relaunched as a single creative platform that runs outside models such as Seedance, GPT Image and MiniMax H3. It says Seedance generations cost half as much on Pika as on other platforms.
Adobe completes Topaz Labs acquisition for AI video and image upscaling
Adobe closed its purchase of Topaz Labs, the Emmy-winning maker of AI upscaling and restoration tools. The technology is coming to Firefly and Photoshop, while Topaz stays a separate brand.
Adobe Premiere mobile arrives on Android for free
Premiere's mobile editor is now free on Android 13 or later. It offers multi-track timelines and AI generative features, with paid add-ons for extra storage and generative credits.
Anil Chakravarthy to replace Shantanu Narayen as Adobe CEO on 1 Dec
Adobe named Anil Chakravarthy, who runs its customer-experience business, as president and CEO from 1 December 2026. Shantanu Narayen will become executive chair.
Character.AI launches Comics that turn chats into comic books
The new (c.ai) Comics feature turns chat conversations into visual comic books. In early data, over 75% of users put themselves in the story and 36% made multi-page comics.
Reve shuts down its image generator on 27 September
Reve stopped new signups on 17 September and ended image generation on 27 September; users can export their libraries until 31 October. It is moving away from creative tools after an OpenAI investment in July.
Figma opens community publishing for Weave AI image tools
Figma users can now publish and share Weave tools, its platform for repeatable AI image workflows. Figma showcased five community tools, including one that turns interior moodboards into 3D renders.
Black Forest Labs releases FLUX 3 Action, an open 7B robotics model
The FLUX image-model maker released an open-weights model for controlling robots. It ranked first on RoboLab-120 with a 42.92% success rate and runs up to 3.95x faster than previous leading systems.
Most active investors
Ranked by how many of this month's deals they joined.
Andreessen Horowitz
4 deals this month, led 1
Lightspeed Venture Partners
4 deals this month, led 1
General Catalyst
2 deals this month
Sequoia
2 deals this month
Also backing this month's deals
Samsung Electronics
CorporateSouth Korean electronics giant that invests in startups.
LedAdvent
Growth equityBoston-based global private equity firm doing large buyouts.
JoinedASML
CorporateDutch maker of the machines that print advanced computer chips.
JoinedBlackRock
Growth equityWorld's largest asset manager, also investing in private companies.
JoinedNvidia
CorporateAI chip giant that invests widely in AI startups.
JoinedPSG Equity
Growth equityBoston growth equity firm backing software and tech-services companies.
JoinedSalesforce Ventures
CorporateSalesforce's venture arm investing in enterprise software and AI.
JoinedScaleup Europe Fund
Growth equityEQT-run €5B EU-backed fund for European tech scaleups.
JoinedAccel
Venture firmEarly and growth-stage venture firm, early backer of Facebook.
Joined
Cognition$2BAvenir
Growth equityNew York growth equity firm making concentrated tech bets.
JoinedCognition$2BFounders Fund
Venture firmSan Francisco venture firm co-founded by Peter Thiel.
JoinedCognition$2BDiffusion
Venture firmNew AI fund from Victor Lazarte and ex-Coatue investor Kris Fredrickson.
Led
Harvey$550MCoatue
Growth equityTech-focused investment firm spanning public stocks and private growth deals.
JoinedHarvey$550MGIC
Growth equitySingapore's sovereign wealth fund investing globally.
JoinedHarvey$550MGoldman Sachs Alternatives
Growth equityGoldman Sachs' private equity, credit and growth investing arm.
Joined
Temporal$550MKleiner Perkins
Venture firmPioneering Menlo Park venture firm, early backer of Google and Amazon.
JoinedHarvey$550MTiger Global
Growth equityNew York investment firm known for fast, large tech growth bets.
JoinedTemporal$550MWellington Management
Growth equityBoston asset manager that also invests in late-stage private companies.
JoinedTemporal$550MWndrCo
Venture firmTech investment firm co-founded by Jeffrey Katzenberg, now branded WNDR.
JoinedHarvey$550M
Insight Partners
Growth equityNew York growth-stage investor focused on software companies.
Led
Snorkel AI$350MAddition
Venture firmNew York venture firm founded by ex-Tiger Global investor Lee Fixel.
JoinedSnorkel AI$350MGreylock
Venture firmSilicon Valley venture firm, early backer of LinkedIn and Airbnb.
JoinedSnorkel AI$350MGV
CorporateAlphabet's venture capital arm, formerly Google Ventures.
JoinedSnorkel AI$350MS32
Venture firmVenture firm founded by Google Ventures founder Bill Maris.
JoinedSnorkel AI$350MWells Fargo
CorporateMajor US bank investing through its strategic capital arm.
JoinedSnorkel AI$350MMultiples Alternate Asset Management
Growth equityMumbai private equity firm founded by Renuka Ramnath.
LedBABrahma AI$150MY Combinator
Seed fundSilicon Valley startup accelerator behind Airbnb, Stripe and Dropbox.
LedDecacorn VC
Venture firmSingapore-based cross-border tech venture fund.
JoinedLobster Capital
Seed fundSmall fund investing only in top Y Combinator startups.
JoinedMento VC
Seed fundPre-seed and seed fund backing AI and future-of-work startups.
JoinedPioneer Fund
Seed fundFund of YC alumni investing in select Y Combinator companies.
JoinedRebel Fund
Seed fundData-driven seed fund that backs only Y Combinator startups.
JoinedStretford End Capital
Venture firmSmall early-stage venture firm named after Manchester United's stand.
JoinedWhat to watch in October
Dates and launches coming up next, and how sure we are about each.
- 29 SeptConfirmed
OpenAI DevDay
Expect GPT-6 API and developer news in San Francisco, and possibly new image and media APIs now that Sora is gone.
- 30 SeptConfirmed
Gemini Omni Flash preview shuts down
Apps still using the preview model have to move to gemini-omni-1.1-flash.
- Coming weeksExpected
Claude Sonnet 5.5 and Haiku 5.5
Anthropic says the smaller models follow Opus 5.5 soon, likely at lower prices.
- Not announcedRumoured
Kling 4.0
Widely expected but not confirmed. A launch would reset the price and quality bar for multi-shot 4K story video.
Stories are summarized in plain language; follow the link before quoting a figure. Story images are shown from the publishers' own sites.
Photos via Wikimedia Commons: Arthur Mensch by Jan2342342423, CC0; Clément Delangue by SiliconANGLE theCUBE, CC BY 3.0; Dario Amodei by TechCrunch, CC BY 2.0; James Cameron by Kevin Paul, CC BY 4.0; Jeff Dean by Cmichel67, CC BY-SA 4.0; Jeffrey Katzenberg by Collision Conf, CC BY 2.0; Sam Altman by Office of the Prime Minister of Japan, CC BY 4.0.