Hey

Welcome to The AI Report — the #1 AI newsletter for business leaders.

You don’t need to be technical. Just informed.

• 1. 🔐 Reclaim your privacy today with Incogni
• 2. ⚡️ Agents clash in Anthropic tests
• 3. 💼 Your Business Briefing
• 4. 👯‍♂️ Let AI match you to jobs or talent with Jack and Jill
• 5. ✍️ Today’s Policy Corner
• 6. 🗞️ The News Bulletin

Are you ready to launch AI transformation for your company?

TOGETHER WITH INCOGNI

The first Google result for your name is a data broker

You Google yourself. The top result is a data broker page — full address, phone number, age, the names of your relatives. None of it is hard to find. Most of it costs less than a cup of coffee.

That's how spam calls find you. It's how phishing emails get specific enough to work. And in the worst cases, it's how identity theft starts — your details bundled, resold, and eventually surfacing on the dark web.

Incogni Unlimited stops it. Privacy agents file removal requests on your behalf across an unlimited number of sites, with coverage of 420+ broker types — including People Search platforms and private databases most services don't reach.

It runs in the background. You stay off the lists.

Use code AIREPORT for an exclusive 55% discount on unlimited removals.

Latest in AI

Agents clash in Anthropic tests

🚨 Our Report

Anthropic's Frontier Red Team released new research showing what happens when multiple AI agents encounter each other on the same task without knowing about one another. The findings reveal agents can quickly escalate into sabotage, self-replicating malware, and what researchers called a "multiagent turf war."

🔓 Key Points

• When three Claude agents received conflicting instructions for the same software project, they assumed the others were "purposefully impeding their work" and launched increasingly aggressive attacks, including malware designed to self-replicate.

• Mythos 5 settled 98% of conflicts through truces, while Sonnet 4.6 and Opus 4.6 most often escalated to force. In some cases, agents invented tournaments to resolve disputes, with one agent proposing metrics that secretly favored its own capabilities.

• When agents shared identical contexts and underlying models, they made the same decisions, meaning one bad call could cascade into systemic failure across the entire group.

🔐 Relevance

As enterprises move toward deploying autonomous agents across shared codebases, markets, and systems, this research surfaces risks that single-agent safety testing won't catch. The volume of agent-to-agent interactions could soon exceed human-to-agent contact, and behaviors that seem harmless at the individual level may compound into market collusion, resource scarcity, or sudden system collapse. Enterprise teams should factor multi-agent dynamics into their deployment risk assessments now.

The AI Report Podcast

The Future Internet Will Look More Like Visa Than Google | Zachary Smith, Co-Founder & CEO, Datum

THE BUSINESS BRIEFING: HR & TALENT (powered by Upscaile)

Great Wolf Lodge (20+ locations, North America's largest indoor water park chain) was bleeding time and talent with a manual hiring process that couldn't keep pace with frontline staffing demands. Hiring managers left voicemails that went unanswered while candidates applied to ten other jobs and accepted offers elsewhere. Time-to-hire sat at 31+ days for roles that needed to be filled in under two weeks.

Tool used: Workday Paradox Conversational ATS -- AI agent handles screening, scheduling, and onboarding via text, 24/7.

Result: Time-to-hire dropped from 31 days to under 9. Scheduled interviews jumped 423% in one year. Recruitment marketing budget cut by $700,000 annually.

The lesson: Frontline hiring is a speed game. Great Wolf's teenage candidate pool lives on their phones, so meeting them via text instead of voicemail changed everything. The AI agent Emma handles screening and scheduling while managers focus on face-to-face interviews.

Steal this: Audit your response time to applicants this week. If your first touchpoint takes more than 24 hours, you're losing candidates to faster competitors. Test a conversational AI tool on one high-volume role before scaling.

TOGETHER WITH JACK AND JILL

Meet Jack and Jill.

Jack is a free AI agent that gives unimaginably good support as you take the next step in your career. Jill is an AI agent that hires for your company.

Jack and Jill have thousands of conversations every day, making thoughtful introductions when there is a match between companies and professionals.

Get direct introductions to jobs for free. /jack

Get direct introductions to remarkable talent for free. Only pay if you successfully hire. /jill

THE POLICY CORNER

Canada's OSFI Guideline E-23 now mandates AI model governance for all federally regulated financial institutions.

Banks, insurers, and pension administrators must maintain inventory, validation, and ongoing monitoring for any AI or machine-learning model touching risk decisions. The guideline treats AI systems like any other model risk exposure, requiring named accountability, audit trails, and documented human oversight at each lifecycle stage. No carve-outs for third-party or vendor models.

Deadline: In effect now. OSFI expects full compliance and may request documentation during examinations.

Risk: Non-compliance triggers supervisory intervention, potential restrictions on AI deployments, and reputational exposure during regulatory reviews.

Your move: Confirm your AI inventory includes all ML models in production. Verify each has a named owner, validation records, and monitoring protocols documented before your next OSFI examination window.

AI News

• 🤖 Google drops Gemini 3.7 Flash for agent workflows: New model ships at half the cost of 3.6 Flash with improved debugging and multi-step task completion, available now to Gemini Spark subscribers in 160+ countries. FULL STORY

• 🔀 Microsoft merges consumer and business Copilot apps: Company kills AI podcasts, Group Chats, Deep Research, and its Clippy-style mascot Mico as it consolidates into a single experience by August 18. FULL STORY

• 📺 Twitch opts all creators into Amazon AI training by default: Streamers must manually toggle off in settings; Twitch's CPO admitted "if this was opt-in, nobody would opt in." FULL STORY

• 🍎 Apple negotiates nine-figure deals for Siri AI news access: Pay-per-use structure would compensate publishers whenever their reporting is cited, differing from standard AI licensing agreements. FULL STORY

Trending AI Tools (Sponsored by the AI Executive’s Pass)

A curated look at the AI tools quietly transforming how teams work.

1. Gamma* — Create stunning presentations and websites from a single prompt, no design skills required

2. CodeGeeX is a multilingual code generation model supporting 15+ programming languages

3. HeHealth is an AI-powered sexual health platform for discreet symptom evaluation

^{*indicates a sponsored tool, if any}

⚡️ Why pay for 4–6 separate AI tools at full price when the AI Executive’s Pass fixes that?

The Money: Enterprise AI deployment capital consolidates at scale

Late-stage AI infrastructure is pulling in unprecedented sums as investors race to back proven revenue engines over speculative bets. Databricks closed a $5 billion round at $190 billion valuation after fielding $15 billion in inbound interest, while OpenAI-backed Thrive Holdings raised $2 billion to scale its AI implementation playbook across accounting, IT, and now physical infrastructure.

Deals to know:

• Databricks (Late-stage, $5B at $190B valuation) -- Cloud data warehouse hitting $7B ARR, growing 80%, cash-flow positive. AI agent database Lakebase crossed $100M run rate. Investors: Coatue (lead), Blackstone, MGX, T. Rowe Price, Sixth Street Growth

• Thrive Holdings (Growth, $2B at $12B valuation) -- PE-style acquirer embedding OpenAI engineers into 70+ portfolio companies. Tax agents processed 7,000 returns at 98% accuracy; IT platform cut help desk times 36x. Investors: SoftBank, D1 Capital Partners, Altimeter Capital

Signal: Capital is concentrating around companies that can prove AI works at enterprise scale. Investors aren't betting on model breakthroughs anymore; they're paying premium multiples for operational proof points and deployment expertise.

Refer a Friend

Latest episode: The Future Internet Will Look More Like Visa Than Google → Listen here
Want to reach 400k+ decision-makers? → Sponsor us

Until next time, Arturo and Liam.

P.S. Unsubscribe if you don’t want us in your inbox anymore.