GlobalDots

Fast cars. Faster conversations. Our Dublin edition of Fast Gear Shifting on Security brought together 10 C-level leaders to dig into Zero Trust and AI in the workplace with Cloudflare. Then we took it to the racetrack. Ferraris included. See you at the next one.

Last night in Berlin was exactly what good industry events should look like. No slides. No pitch deck. Just practitioners talking honestly about what AI is doing to cloud spend. Thanks to CloudHiro for co-hosting with us, and to everyone who showed up for the conversation. We're still in town for #WeAreDevelopers. Come say hello :)

Disesdi Shoshana Cox didn't set out to challenge the AI security industry. She first raised her concerns privately. She says the response was: "We don't care as long as people pay for it." Speaking out, she says, eventually led to threats and strengthened her resolve to explain why she believes today's AI Red Teaming creates false confidence rather than meaningful security. In this clip from our CloudNext Podcast, she explains to Ganesh The Awesome why today's approach to AI Red Teaming is fundamentally flawed. Check out the full episode in the comments.

Most security teams we talk to have an acceptable-use policy for AI tools. Almost none have network-level enforcement behind them.   No DLP enforcement on outbound prompts. No control over which models are queried. No audit trail.   This isn't a behavior problem. It's an enforcement gap.   On July 29, GlobalDots is running a 30-minute session on closing it. Zero Trust. SASE. AI access governance. Live demo.

Last week, our team was at GITEX Europe in Berlin. Great conversations, great energy. This week, we're back at it! We'll be at WeAreDevelopers World Congress, and on Wednesday evening we're hosting a FinOps networking dinner: Cloud Cost Cocktails, together with CloudHiro. If you're around, join us. RSVP for the dinner (Link in the comments)

Do you know every AI agent talking to your systems right now? Not employees pasting data into ChatGPT. Agents calling other agents. Calling MCP servers. Calling internal APIs. Yesterday at Cato Networks' offices, our own Dror Arie ran a hands-on workshop and roundtable with Zebra Technologies on exactly that. One line from the room stuck: over-permissioned agents are the new over-permissioned service accounts. Prompt injection turns them into a confused deputy, acting on an attacker's behalf. No slides-only session. Real scenarios, worked through in the room. More of these are coming.

Edge performance without guesswork. Security testing before attackers do. That's what we're bringing to Berlin. Our team will be there tomorrow and Tuesday at GITEX Europe (Hall 4.2, Booth B10) alongside our great partners Bugcrowd and Hydrolix. If you're working on edge performance or security, come say hello.

"You can spray a bunch of prompts at something and say: 'Look, all these attacks worked, I blocked them all now.' But did you actually provide meaningful security? I would argue no." New CloudNext episode with Disesdi Shoshana Cox on why AI red teaming doesn't work and what does.

Your employees are already using AI. Do you know which ones? Shadow AI is the new shadow IT and most security teams are still catching up. June 30, Tel Aviv. Hands-on workshop with Cato Networks + Zebra Technologies. Limited seats.

Ferrari on a racetrack. Zero Trust in the boardroom. A small group of security leaders. One closed roundtable. No slides. Details and registration below | July 8, Dublin Edition