The Italian cybersecurity startup, Exein, has raised $270 million in a funding round, reaching unicorn status. The company is focused on combating cyber threats by leveraging advanced AI-powered technologies to counteract AI-driven hackers.
Microsoft has introduced a draft of its AI code of conduct, outlining principles for what its AI models should and shouldn’t do. The company is seeking public input before releasing a finalized version later this year.
A recent data breach at financial technology company Revolut has affected a total of 680 account holders worldwide, including 12 users in Ireland. The leaked information includes sensitive details such as passport and bank data.
The complex 3D models used in building projects today are powerful but often difficult to navigate, leading to costly mistakes when engineers, consultants, and contractors handle them. Autodesk, a leading software company in architecture and construction modeling, has introduced Autodesk Assistant, an AI-powered tool designed to simplify this process. Announced at Autodesk University, this assistant automates and streamlines building design workflows, allowing multiple stakeholders to collaborate effortlessly without needing to master complex software. Using a geometry-based AI model called Neural CAD, Autodesk Assistant can answer queries, adjust designs in real-time, track construction progress, and manage operations—all in a single dynamic application. For example, it can instantly reshape floor plans, flag issues like the need for new hardware, and communicate project updates to the team, enabling quicker decision-making and reducing dependency on specialized software. This tool aims to allow architects, engineers, and builders to focus more on solving design problems rather than struggling with software intricacies, fostering greater collaboration and higher precision in project outcomes.
During a San Francisco conference, Anthropic's CEO Dario Amodei called for a cautious slowdown in AI progress, using a car safety analogy to stress the importance of responsible development after incidents with competitors. Shortly after, Nvidia's CEO Jensen Huang expressed opposition to slowing down AI innovation. Amodei emphasized the need for self-reflection and shared accountability rather than blaming others in the race for AI advancement.
In 2026, U.S. venture-backed tech companies have raised nearly $90 billion through public offerings, ranking as one of the highest on record. However, the vast majority of this capital was secured by just two heavyweights: SpaceX, capturing 83% of the funds, and Cerebras Systems, with 6%. The rest of the tech IPO landscape remains modest, with only 21 other sizable offerings including a mix of traditional IPOs and SPACs totaling less than $10 billion. Notably absent this year are enterprise software IPOs—a sector traditionally represented among tech offerings—which may be shifting focus towards AI-driven innovations instead. Meanwhile, sectors like energy, defense, aerospace, medical devices, and consumer products have seen relatively strong activity. This year saw energy startups, especially in geothermal and nuclear technologies, leading the IPO count, quantum computing and defense tech making solid entries, and consumer mobility platforms like Lime debuting despite valuation challenges. The uneven distribution highlights a winner-takes-almost-all dynamic in tech IPOs, with big players dominating returns and smaller companies holding back. Pipeline IPOs similarly are dominated by giants like Anthropic and OpenAI, leaving enterprise software offerings largely absent from the immediate horizon.
Anthropic and OpenAI plan to integrate independent safety evaluators within their AI research teams. While this move is praised for offering rare access to internal processes, experts caution that effective oversight hinges on genuine independence, transparency, and the introduction of formal regulations.
Apple is reportedly working on enterprise servers using its custom M8 Ultra chips. They have explored integrating Nvidia's NVLink Fusion technology to link these chips, targeting AI developers, corporations, and government entities for deployment around 2029. This development aligns with the European Commission's recent Cloud and AI Development Act, which evaluates public cloud contracts based on their advancements in cloud and AI technologies. The move signals Apple's ambition to enter the AI infrastructure market with a high-performance, chip-based server solution.
Mark McQuade, co-founder and CEO of Arcee AI, emphasized that organizations shouldn't have to compromise between using cutting-edge AI models and maintaining control over the technology critical to their operations. San Francisco-based Arcee AI specializes in creating open-weight AI models and has recently completed a Series B funding round that has pushed its valuation over the $1 billion mark.
OWASP’s 2026 Top 10 list for large language model (LLM) applications has shifted its focus, raising 'excessive agency' from sixth to third position and moving 'improper output handling' down from fifth to tenth, reflecting insights drawn from recent incident data. Meanwhile, Europe’s new Cyber Resilience Act, effective from 11 September, mandates that product vulnerabilities be reported within 24 hours. However, this legislation targets product security rather than the specific ways AI agents are deployed. These developments underscore the evolving landscape of AI security regulations and highlight the pressing challenges in managing agent permissions and outputs.
