Welcome, AI Insiders!

One founder, no VCs, six months. A solo dev just flipped an AI app for $80M. OpenAI raised the alarm on future models crossing biosecurity lines. Stanford revealed that workers want AI to assist, not replace. And beneath it all, a quiet shift: the frontier of AI is no longer just tech

In today’s Generative AI Newsletter:

  • Wix acquires Base44 for $80M in a landmark solo AI exit

  • Stanford reveals workers want AI collaboration, not replacement

  • OpenAI warns its next models may exceed biosecurity risk thresholds

Special highlight from our network

𝐓𝐮𝐦𝐞𝐫𝐲𝐤’𝐬 𝐥𝐚𝐭𝐞𝐬𝐭 𝐒𝐭𝐚𝐭𝐞 𝐨𝐟 𝐀𝐈 𝐓𝐫𝐮𝐬𝐭 𝐑𝐞𝐩𝐨𝐫𝐭 𝐢𝐬 𝐡𝐞𝐫𝐞 𝐚𝐧𝐝 𝐢𝐭 𝐜𝐡𝐚𝐧𝐠𝐞𝐬 𝐭𝐡𝐞 𝐰𝐚𝐲 𝐰𝐞 𝐦𝐞𝐚𝐬𝐮𝐫𝐞 𝐆𝐞𝐧𝐀𝐈 𝐫𝐞𝐚𝐝𝐢𝐧𝐞𝐬𝐬.

We’ve analyzed leading LLMs including Claude, Gemini, DeepSeek and OpenAI o4-mini, across nine critical dimensions of risk.

👉 Claude 4 scores lower than Claude 3.7 on trust

👉 The top performer isn’t who you think

👉 Hallucinations and output issues still plague even the best models

👉 From prompt injection gaps to fairness scores — this report has it all

𝐃𝐨𝐰𝐧𝐥𝐨𝐚𝐝 𝐭𝐡𝐞 𝐟𝐮𝐥𝐥 𝐑𝐞𝐩𝐨𝐫𝐭 𝐧𝐨𝐰

Want to know how we score trust in AI?

Special highlight from our network

Why AI agents are hitting a data security wall — and how to break through

Everyone’s excited about AI agents. But at many enterprises, deployments are stalling.
Not because of the tech — because of data security.

🎯 Join Steve Nouri and Amar Kanagaraj, Founder & CEO of Protecto, for a live conversation on tackling this critical challenge. Amar works closely with Fortune 500 companies, banks, and healthcare organizations to solve exactly this problem — sharing insights you won’t find anywhere else.

📅 When: June 27
Time: 3 PM CET

This event is designed for enterprise leaders, security pros, and anyone building AI-powered products who wants to understand what it really takes to deploy AI at scale.

Special highlight from our network

Ready to meet the minds shaping the future of AI?

From OpenAI to Llama3, from startup legends to global builders, the biggest AI event of the year is your chance to learn from the very best.

🎤 Steve Nouri (GenAI.Works)
🎤 Jason Wei (OpenAI)
🎤 Arvind Jain (Glean)
🎤 Soumya Batra (Llama2 & Llama3)

🎟️ Get 15% off with code: GENAICOMMUNITY

🚀 AI Solo Project Sells for $80M: A Milestone for Vibe Coding

Image Credit: Forbes

In one of the most stunning AI startup exits of the year, developer Maor Shlomo has sold his AI-native app builder Base44 to Wix for $80M in cash. Built without investors, the “vibe coding” platform used LLMs to turn plain prompts into full apps, racking up 250K users and $189K monthly profit in just six months.

Key happenings:

  • Base44 let non-coders build entire apps via text, with built-in database, auth, analytics, maps, and messaging

  • Launched in January, reached 10K users in 3 weeks via word-of-mouth and social media posts by Shlomo

  • Profitable by May despite LLM token costs; landed clients like eToro, Similarweb, and demoed at an AWS event after switching from OpenAI to Claude via AWS

  • Shlomo owned 100%, with $25M in employee bonuses baked into the deal; called Wix the “best possible partner” to scale the product

Base44 is one of the first serious proofs that LLMs can power solo-run, no-funding startups to multi-million dollar exits. Shlomo’s rapid build, profitable scale, and lightning-speed acquisition show how much AI is compressing the startup timeline and redefining who gets to build, launch, and win in tech.

🧠 Stanford Study Reveals What Workers Really Want From AI

Image Credit: Stanford

Stanford surveyed 1,500 workers across 104 occupations to understand how people actually want AI agents involved in their jobs. The study uncovered a striking mismatch between current AI investments and worker preferences, with most people asking for help with boring tasks, not full automation.

Key Findings:

  • 41% of YC AI startups target tasks workers don’t want automated

  • Top request: automate repetitive, low-value jobs like scheduling and data entry

  • Arts/media workers showed strongest resistance to AI, with only 17% of creative tasks rated suitable for automation

  • Nearly half of all occupations prefer equal human-AI partnership over full automation, according to the new “Human Agency Scale”

While much of Silicon Valley chases full automation, workers are asking for collaboration. It’s shaping real inefficiencies in where AI talent and capital are flowing. As agents become central to how work gets done, respecting human preferences may be the key to unlocking true productivity gains.

🧬 OpenAI Warns Its Next Models May Break Biosecurity Thresholds

Image Credit: Getty Images

OpenAI just issued a public warning: its future models may soon be powerful enough to aid in the creation of biological weapons, crossing into “high risk” territory under its internal Preparedness Framework.

Key details:

  • Future successors to o3 are expected to meet the threshold for dangerous biological capability, prompting OpenAI to activate a new set of safeguards

  • Mitigations include refusal training, constant monitoring, human-in-the-loop reviews, and red teaming with biology experts

  • A biodefense summit will be held in July with governments and NGOs to coordinate countermeasures and advance biosecurity research

he company is proactively tightening controls while lobbying for broader defenses, including diagnostics, synthesis screening, and early pathogen detection. It’s a major pivot toward biosecurity as a core part of AI safety infrastructure and a sign that the biotech risks from AI are now on the industry's front burner.

🚀 Boost your business with us—advertise where 10M+ AI leaders engage

🌟 Sign up for the first AI Hub in the world.

Reply

or to participate

Keep Reading

No posts found