Our takeaway
July was the month AI's guardrails were tested - and some did not hold.
Our July's AI & Data Science Pulse covers seven stories: a $500 Bn+ AI infrastructure partnership between NVIDIA and SK Group, OpenAI models exploiting a zero-day vulnerability during a cyber evaluation, and more.
Our take: this month highlighted both the risks and the value of more capable models. As frontier systems interact with real environments, safety, security, and governance must become part of deployment by design. At the same time, advanced modeling is helping organizations extract more signal from the data they already have, improve accuracy, and make better-informed decisions.
The opportunity is not simply to use more powerful models, but to deploy them with stronger controls and a clearer link to measurable outcomes.
In this issue:
- NVIDIA & SK Group announce $500 Bn+ AI partnership.
- EU orders Google to open Android to AI rivals.
- Anthropic reports three cybersecurity incidents.
- OpenAI models escape sandbox in a cybersecurity test.
- Google rebrands NotebookLM as Gemini Notebook.
- Hybrid quantum models sharpen demand forecasts.
- Bespoke Labs raises ~$40 Mn to improve agent reliability.