The AI Highlights Review brings together the content recently featured at Data Phoenix. This edition opens with a brief recollection of all the ways in which OpenAI has managed to take over the news cycle, ranging from new model releases to a massive funding round announcement to faulty charts.

OpenAI started its rollercoaster of a week by announcing it had raised an astonishing $8.3 billion at a $300 billion valuation in a funding round that reached such demand that early investors had to be granted smaller allocations so new strategic investors could participate. The round is said to be part of the company's ambitious plans to raise $40 billion by the end of the year.

Crucially, OpenAI has to convert to a for-profit structure for that plan to work. This is because SoftBank, which is footing 75% of the total investment, can cut back on its investment if OpenAI fails to convert to a for-profit structure by December 31. One of the hurdles OpenAI must clear is that Microsoft, its most important backer, must approve the conversion plan.

Recent reports by several outlets mention that the two companies have been in talks to reach an agreement, and that OpenAI may have to agree to continue providing preferential access to Microsoft even after the former fulfills its theoretical goal of reaching artificial general intelligence, or AGI. A previous agreement between the two companies established that OpenAI would no longer be obliged to give Microsoft access to its intellectual property once the startup successfully developed what it internally defines as AGI.

It is very likely that the news of a potential agreement, in addition to reports that OpenAI had reached 700 million weekly active users and was well underway to reach $20 billion annualized revenue by the end of the year, all contributed to the round's popularity. Further excitement about the startup's future was fueled by news that OpenAI is in talks to sell some shares held by current and former employees. The share sale would value the company at half a trillion dollars, effectively surpassing Elon Musk's SpaceX, currently valued at $350 million, and expected to rise to $400 million once it completes its most recent funding round.

On Tuesday, the company dropped two open-weight models after experiencing some delays. The much-anticipated reasoning models are OpenAI's first open offering since GPT-2 in 2019. The larger model, gpt-oss-120b, runs on a single Nvidia GPU, making it ideal for use in data centers and high-end consumer hardware. Meanwhile, the lighter gpt-oss-20b can be run on consumer laptops with at least 16GB of memory.

Finally, on Thursday, the even more highly anticipated GPT-5 was launched. OpenAI famously faced several setbacks in the development of this model, and the constant delays contributed to the eventual realization that the so-called "AI scaling laws", which broadly state that model performance scales with increases in data and compute, were reaching a point of diminishing returns. Indeed, many have already pointed out that although GPT-5 is an improvement over previous OpenAI models, it certainly wasn't the breakthrough that some expected it would be.

Other noteworthy recent headlines

Swedish AI coding startup Lovable becomes a unicorn and hits $100M ARR in just 8 months: Swedish AI coding startup Lovable became Europe's fastest-growing unicorn in just eight months, raising $200M at a $1.8B valuation. Additionally, Lovable has reached $100M ARR, largely due to its 2.3 million users who have created over 10 million projects using its no-code platform.

Latent Labs' new model assists protein binder design through a no-code web interface: Latent Labs has made its new frontier model for binding generation, Latent-X, available through a no-code web platform. With Latent-X, Latent Labs aims to accelerate the costly drug design process by drastically shortening the time needed to identify potential binders for novel proteins.

AI Compliance Startup Delve Raises $32M at $300M Valuation: AI-powered compliance startup Delve has raised a $32 million Series A at a $300 million valuation. After pivoting from healthcare AI to automate regulatory compliance workflows, the startup now serves over 500 companies, including AI unicorn Lovable.

FuriosaAI turns to LG AI Research after rejecting Meta's $800M offer: FuriosaAI secured a major partnership with LG AI Research to power their EXAONE AI models with superior energy-efficient chips, just months after rejecting Meta's $800 million acquisition offer.

Google DeepMind's Aeneas helps historians restore, date and attribute provenance to Roman inscriptions: Google DeepMind's Aeneas AI model revolutionizes ancient history research by analyzing Roman inscriptions with 73% accuracy in text restoration and helping historians find contextual parallels in seconds rather than weeks.

IdentifAI raises €5M to combat deepfakes as AI-generated content floods the Internet: Italian startup IdentifAI raised €5 million in Series A funding to expand its real-time deepfakes and synthetic content detection platform. IdentifAI's mission has become critical in a world where the explosive growth of AI continues to erode our trust in digital content.

Julius AI raises $10M to create data visualizations from natural language prompts: Julius AI has raised $10 million in seed funding led by Bessemer Venture Partners to expand its natural language data analysis platform. Julius AI can generate insights and visualizations by simply asking questions in plain English.

Google looks to join the "vibe-coding" revolution with Opal, a no-code AI app builder: Google Labs launched Opal, an experimental no-code platform that enables users to build and share AI mini-applications by describing workflows in natural language, which the tool then converts into powerful AI-powered applications.

Fundamental Research Labs secures $33M to build next-gen AI agents: Fundamental Research Labs raised $33M in Series A funding to develop autonomous AI agents that bring cutting-edge research to consumer products designed to work alongside humans. The company's products include a spreadsheet-based agent, "Shortcuts", and a general-purpose assistant, "Fairies".

Clay raised a $100M Series C round to continue fueling its GTM engineering revolution: Clay raised $100 million in Series C funding at a $3.1 billion valuation, as the company remains on track to bring in $100 million in revenue. The raised funding will be allocated to accelerate product development, as the company continues developing its "IDE for GTM".