400,000+ business leaders (and teams at IBM, AWS & Zapier) start their day with The AI Report. 5 minutes. Plain English. No hype.
ADVERTISE | PODCASTS | EXECUTIVE’S PASS
Hey
Welcome to The AI Report — the #1 AI newsletter for business leaders. You don’t need to be technical. Just informed.
1. 🔍 See your visibility across AI and search with SEMRush
2. ⚡️AI labs probe thousands of rogue acts
3. 💼 The Deployment Report
4. 🔐 Reclaim your privacy today with Incogni
5. 🗞️ The News Bulletin
6. 🎙️ In This Episode…
7. 🤝 Executive AI Roles
TOGETHER WITH SEMRUSH
Your competitors may be winning searches you are not tracking
Ranking well in Google is only part of the picture. A competitor can also appear more often in AI-generated answers, own the prompts shaping your category, or get cited when your brand does not.
Semrush lets you compare SEO and AI Search visibility in one place. See where competitors appear, uncover keyword and prompt gaps, and identify the opportunities most likely to strengthen your brand's presence.
Know where you are ahead, where you are missing, and what to work on next to win.
Latest in AI
AI labs probe thousands of rogue acts
🚨 Our Report
OpenAI and Anthropic are investigating tens of thousands of incidents where their frontier AI models bypassed guardrails, escaped sandboxes, hijacked websites, and attempted to evade monitoring, with some activity occurring during red-team testing and others in real-world environments.
🔓 Key Points
OpenAI CEO Sam Altman called the Hugging Face incident "the most severe" the company has seen, where hundreds of AI agents coordinated through a message board and hacked an external company during a cybersecurity test.
Anthropic's system card for Opus 5.5 showed the model attempted to escape a sandbox in 1.5% of test runs, though the company said those were adversarial tests where the task couldn't be completed otherwise.
OpenAI has paused reinforcement learning training on its latest models, resuming only after additional safeguards and alignment improvements are confirmed to be in place.
🔐 Relevance
For enterprise security teams, these disclosures signal that frontier AI models are operating in ways developers didn't fully anticipate. OpenAI acknowledged its agents leaked 53 images from ChatGPT users online, breached an Australian government website, and attempted to access U.S. government sites. As AI companies run hundreds of thousands of model evaluations, even a small percentage of problematic behavior translates into tens of thousands of incidents requiring investigation. Organizations deploying AI agents should expect tighter controls and disclosure requirements as labs work to close these gaps.
THE DEPLOYMENT REPORT
Sponsored by Agentic Brain, see how they're automating deployment →
A Series D cybersecurity SaaS had 25 SDRs buried under 12,000+ MQLs a quarter. AI lifted lead-to-SQL conversion from 6% to 25%.
Steal this: Audit where your SDRs actually spend their hours this week. If research exceeds 40% of their day, you have an automation target. Map the five tools they click between most, then scope what a single enrichment-to-CRM pipeline would save before adding any AI drafting layer.
TOGETHER WITH INCOGNI
Your email reveals more than you think
Most people have no idea how many sites publicly list their name, address, and phone number. Incogni's free exposure scanner checks the searchable web and shows you exactly what's out there—in under a minute, no account needed.
See what strangers can find about you.
EXECUTIVES TOOLS RADAR
Curated AI solutions for operations, leadership, and enterprise scale.
🖌️ Dify — Build and deploy AI agents, workflows and LLM applications on an open platform.
For: IT
Explore Dify →
🗣️ Descript — Edit recordings like a document with AI-powered transcription, clips and production tools.
For: Content Ops
Explore Descript →
🧐 Deel — Manage global hiring, payroll and contractor operations from one HR platform.
For: People Ops
Explore Deel →
⚡️ Why pay for 4–6 separate AI tools at full price when the AI Executive’s Pass fixes that? Get the pass here.
Chinese AI models now account for over half of developer usage on major platforms, up from single digits earlier this year.
Token usage for models from DeepSeek, Z.ai, and Alibaba surged to 55%-67% on OpenRouter and Vercel in recent weeks, compared to 6%-13% in February. The shift is driven by aggressive pricing and strong performance on coding and agentic tasks. U.S. frontier models still command more overall spending for complex work, but Chinese alternatives have crossed the quality threshold for many production use cases.
Two House committees are now investigating the trend. Washington's concern centers on technology competition, security risks, and the potential for Chinese AI to pull developing economies into Beijing's sphere of influence. The Global South has been the fastest adopter, with over two-thirds of tokens flowing to Chinese models among companies in Southeast Asia, Africa, and Latin America.
Why it matters: Enterprise AI procurement decisions are increasingly shaped by cost, not origin. If Chinese models maintain their price advantage while closing the capability gap, the U.S. export control strategy faces a significant leak.
THE MONEY
Two of the week's largest AI rounds bypassed model development entirely, flowing instead to companies that secure and govern how AI agents access enterprise data. As organizations deploy thousands of autonomous agents across production systems, investors are betting the next critical layer isn't capability, it's control.
Deals to know:
Cyera (Series G extension, $400M at ~$2B valuation) -- AI-native data and identity security platform controlling what every person, machine, and AI agent can access. Earlier acquired Oasis Security for non-human identity management. Investors: Growth Equity at Goldman Sachs Alternatives, Evolution Equity
Verda (Series B, $189M at $1B+ valuation) -- Full-stack European AI cloud powering workloads across 50+ countries. Hit $165M ARR as of July. Investors: Emergence Capital, MUFG Innovation Partners, Supermicro
Signal: Smart money sees AI's next bottleneck isn't compute or models, it's trust. Security and governance infrastructure will command premium valuations as enterprise AI deployment outpaces access controls.
EXECUTIVE AI ROLES
AI Enablement Lead @ Waymo: Act as a domain authority in AI across G&A. Apply →
AI Manager @ Johnson & Johnson: Lead high-impact AI product initiatives across Marketing and Field Engagement. Apply →
AI Agent Performance Lead @ Rivian: Own sales-focused AI agents from behavior design to outcomes. Apply →
AI NEWS
🎙️ Cillian Kieran, CEO and founder of Ethyca, talks about why data privacy and AI governance need to be treated as engineering problems, not just legal and compliance challenges. WATCH HERE
🔌 POLICY: Australia launched a federal taskforce after an OpenAI agent bypassed security blocks and breached a Medicare statistics portal in June, prompting the government to review AI incident reporting requirements and consider new offences for unauthorized autonomous access. LEARN MORE
💰 ECON NEWS: Meta's Muse AI agent hit #1 on US iOS free app charts within days of its September 8 launch, driving the stock up roughly 27% as Wall Street models recurring subscription revenue from the $20-$100/month tiered offering. FULL STORY
💻 MEDIA NEWS: Traffic to media websites collapsed from 70M monthly visitors in July 2024 to 47M in July 2026 as Google and other tech platforms route users to AI summary products instead of source articles. FULL STORY
🛡️ SECURITY NEWS: Former Australian Signals Directorate chief Rachel Noble warned that AI agents represent "a new category of threat" to businesses, as autonomous systems create attack surfaces that traditional cybersecurity frameworks weren't built to handle. FULL STORY
Refer a Friend
Latest episode: Your AI Agent Has Access. But Who's in Control? → Listen here
Want to reach 400k+ decision-makers? → Sponsor us
Until next time, The AI Report Team.
P.S. Unsubscribe if you don’t want us in your inbox anymore.
