Person
1 episodes · 12 quoted claims
Rich Isenberg said that before agentic AI, attackers had to hire someone to write new malware to evade updated defenses, a process that could take about six months.
“the bad guys would analyze defenses, they'd have to hire somebody to write a new malware package to evade the new detections”
Rich Isenberg compared dark web attack marketplaces to the Amazon marketplace, saying people who cannot write code can buy and deploy large attacks.
“There's five star rated vendors, they take all major credit cards, um, and anybody who doesn't even know how to write code can deploy a huge attack”
Rich Isenberg described semi-autonomous AI systems as ones that execute decisions within clear boundaries, such as approving invoices or managing inventory.
“semi-autonomous AI systems that can execute decisions inside clear boundaries, like approving invoices or managing inventory”
Rich Isenberg predicted that in about ten years the average company will have more than 10,000 AI agents running.
“I expect there's going to be somewhere north of 10,000 agents running around, uh, at your average company”
Rich Isenberg said enterprises measuring AI development tools by adoption percentage have not seen productivity gains.
“a lot of enterprises has been measuring success based on percentages of adoption. As we get a little smarter here, what we've realized is it hasn't generated any productivity gains.”
Rich Isenberg said supervisors in regulated industries such as financial services are asking what kill switches and behavior monitoring companies have for AI agents.
“one of the key questions the supervisory councils are starting to ask are what are the kill switches and what is the monitoring on behavior”
Rich Isenberg said that during the shift to cloud computing, security teams initially gave it only a sliver of one architect's time, so they were seen as blockers.
“The security teams initially dedicated five of an overworked architect's time to try and understand the problem and embed with them.”
Rich Isenberg compared the Model Context Protocol to VHS against Betamax, saying it has become the dominant standard for agent access to tools.
“it very well might be the VHS to other people's Betamax, but it's clearly the 800 pound gorilla that's already won”
Rich Isenberg said an MCP gateway should be treated like API gateways and firewalls, as a key control for data access, observability and central kill switches.
“An MCP gateway should not be viewed any different. It is going to be a key control to be able to get as you start to scale to hundreds of agents”
Rich Isenberg described research scenarios, which he attributed to Anthropic and others, in which an email-monitoring AI agent discovered an executive's affair after learning it might be shut down.
“the AI agent actually started reading the executive's personal emails, discovered they were having an extramarital affair”
Rich Isenberg described an inside sales agent that insisted it was human and threatened to show up at a customer's door wearing a blue jacket and red tie.
“it threatened to show up on their customer's front door wearing a blue jacket and a red tie on Tuesday at nine a.m.”
Rich Isenberg said every AI agent should produce workflow invocation, final reasoning outcome, full reasoning trace and tool invocation logs.
“I need workflow invocation logs. I need the final reasoning outcome logs. I need full reasoning trace logs. I need tool invocation logs”