Dear Readers,
My friends are launching a cool new AI video product called Vireel today on Product Hunt. Please check it out and give them a vote.
Anyway, yesterday was the big moment - OpenAI released GPT-5. Lightning fast, more precise in context understanding, with impressive new capabilities. But the euphoria is not unanimous: while some are raving about “magical answers”, others only see a solid update without the hoped-for big hit. The picture is mixed - and therein lies the exciting question: are we facing a gentle evolution or a quiet revolution that will only show its full potential in the coming weeks?
In this issue, we take a deep dive into the initial reactions to GPT-5, look at Google's catching-up competition and analyze how the race for the best AI models is intensifying. Plus: why the link between AI and nuclear decision-making is worrying experts, how the battle over global AI regulation between the US and EU is escalating - and the role digital sensory overload plays in our democracies. A packed update that shows: The AI world is not standing still - and anyone who wants to have their say should read on now.
In Today’s Issue:
GPT-5 is finally here, and the AI community is officially divided
Google's AI wants to be your new teaching assistant
The world's most powerful AI just landed in your code editor
OpenAI is giving its best enterprise AI to the US government for just $1
And more AI goodness…
All the best,
GPT-5 is live – but the community is divided
The Takeaway
👉 The honeymoon phase is over: Even GPT-5 is no longer being celebrated unconditionally – the AI community has learned to ask critical questions and set realistic expectations.
👉 Performance alone is no longer enough: Despite impressive speed and capabilities, developers are missing the hoped-for unification of all modalities in a single model.
👉 Competition is intensifying: Google's rise in prediction markets signals real competition – OpenAI can no longer rely on market dominance.
👉 Evaluation standards are rising exponentially: What would have been considered revolutionary a year ago is now perceived as incremental improvement – a sign of the rapid development of the industry.
The day the entire AI community has been waiting for months has finally arrived: GPT-5 was officially launched yesterday by OpenAI. But while some are raving about the “really good vibes” of the new model, others are surprisingly cautious.
The initial reactions paint a fascinating picture: Enthusiastic reviews praise the “lightning fast” performance with complex queries and the uncanny ability to sense what you want to ask next. At the same time, a certain disillusionment is spreading among the developer community. Many had hoped that GPT-5 would finally unite all models – reasoning, image and video generation, voice – “one model to rule them all,” but this expectation has not been met. At the same time, there are repeated references to poor “routing” (i.e., GPT-5's automatic selection between reasoning and non-reasoning), which often does not work and selects the non-reasoning model even for difficult questions.
Particularly exciting: prediction markets are already showing skepticism – Google overtook OpenAI in bets for the best model at the end of August. These mixed reactions are invaluable to our community because they show that we are in a time when the AI landscape is more competitive than ever before.
What will this polarization mean for the future of AI development? Will the skeptical voices be proven right, or will GPT-5 reveal its true strength in the coming weeks?
So far, the overall picture is extremely mixed. GPT-5 seems more like an iterative update than the revolution we had hoped for. It has been improved in many areas, but is underwhelming in key areas.
Why it matters: GPT-5 marks a turning point in AI development, where even groundbreaking releases no longer automatically trigger waves of enthusiasm. The critical voices in the community show a new maturity in dealing with AI promises and set the bar higher than ever for future developments.
Sources:
Kim’s Favorites Things: Vireel.com
Dear Readers,
A few friends of mine are launching a product called Vireel today on Product Hunt.
I tried it out last week and wanted to share, because it’s doing something I haven’t really seen before…
Just give it your website, and it auto-generates short-form videos you can use as ads or organic posts.
But the special thing is it’s all based on proven viral reels and they aren’t just talking avatar vids, but fully polished reels— ready to post.
Please check it out and give them a vote! https://vireel.com/vote
Cheers, Kim
PS. You can use code LAUNCH40 for 40% off for life.
In The News
Google Touts AI for Educators
Google for Education is highlighting how its NotebookLM tool can save educators and administrators hours by instantly sorting through mountains of research and documents to extract key takeaways.
GPT-5 Arrives in VS Code
OpenAI's most powerful model, GPT-5, is now rolling out to all paid GitHub Copilot users within Visual Studio Code, bringing significant advances in reasoning, coding, and chat capabilities.
OpenAI's $1 Offer to Feds
In a major push to deepen its government ties, OpenAI is now offering its powerful ChatGPT Enterprise product to all U.S. federal agencies for just $1 for the next year.
Graph of the Day
The graph of the day shows a literally strange picture: during the livestream, the following “SWE-verified benchmark” was shown, and for some inexplicable reason, it displays the metrics completely incorrectly.
The heights of the corresponding diagram do not accurately reflect the percentage values. It is questionable how such a serious error could occur during a livestream.
Nuclear decision-makers and AI: A risky combination
Experts warn that AI will inevitably find its way into control systems for nuclear weapons – especially in data-based decision-making tools for heads of state. Despite consensus on central human control, there is growing concern that black box systems will distort judgment and that historical mistakes (e.g., the Petrov incident) could be repeated by AI in the future. In the long term, there is a danger of a dangerous blurring of democratic control and technological dominance.
USA vs. EU: Competition over AI regulation escalates
US technology advisor Michael Kratsios urges Asia not to be intimidated by restrictive EU AI regulations. Instead, Washington is focusing on flexible, innovation-friendly approaches, coupled with US chips, data centers, and language models. In doing so, it is driving a geopolitical divide that links democratic security with flexible governance and technological dominance.
Digital information overload – democracy under constant fire
Rafael Behr analyzes how digital noise – accelerated by AI and platforms – is putting democratic institutions under constant cognitive stress. The new UK Online Safety Act is seen as a necessary, albeit imperfect, step: it curbs disinformation, but raises questions about overregulation and potential political instrumentalization. Significantly, democracy faces a systemic conflict between analog structures and digital reality.
Put Your AI & Politics Insight in Front of 200,000+ Readers
Exploring how AI is reshaping politics, policy, or governance? We’re featuring sharp analysis and research on the intersection of AI and political systems in Superintelligence, read by 200k+ people.
If you have significant comments on AI and politics (unemployment, geopolitics, proposed legislation) and would like to have this discussion presented in the newsletter, get in touch via a brief summary to [email protected] with the subject line “Politics Submission”. We’ll reach out if your work is a fit for a future issue.
Question of the Day
GPT-5: success or disappointment?
Tipp of the Day
OpenAI released their GPT-5 promting guide. Give it a try:
Tweet of the Day
GPT-5 now rolled out to 20% of paid users and doing >2B TPM on the API! so far so good...
excellent work by the eng and infra teams!
— #Sam Altman (#@sama)
9:07 PM • Aug 7, 2025
GPT-5 is slowly rolling out to everyone.
Sponsored By Vireel.com
Vireel is the easiest way to get thousands or even millions of eyeballs on your product. Generate 100's of ads from proven formulas in minutes. It’s like having an army of influencers in your pocket, starting at just $3 per viral video.
Rumours, Leaks, and Dustups
K2 benchmark - GPT-5 version
— #Crystal (#@crystalsssup)
6:48 AM • Aug 8, 2025
Numerous people are making fun of OpenAI's embarrassment during the livestream regarding the charts. Rightly so, in our opinion.
GPT 5 is dissapointing ngl
— #Dylan Patel (#@dylan522p)
2:36 AM • Aug 8, 2025
The current mood seems to be dominated by disappointment. GPT-5 does not appear to be delivering the expected improvements for many users.