Key facts
- Home Depot is in active conversations with Meta about how its Muse agent should identify itself to retail websites, CIO Angie Brown said.
- Amazon barred Meta's Muse from its marketplace, saying the bots did not properly identify themselves.
- Walmart, Best Buy, Sephora and Shopify have announced integrations or openness to AI agents, according to Business Insider.
- Brown said Home Depot's main website was not affected by the PHP vulnerability wave and a few smaller units patched flaws.
- Home Depot has access to Anthropic's Mythos model but was not granted early access via Project Glasswing, Brown said.
Home Depot wants to let AI agents search for and buy products on a customer's behalf, but only if the agent can prove it is acting for a real person, Angie Brown, the retailer's executive vice president and chief information officer, told Business Insider in an interview.
"We want to know you're not a bot, and we actually want to know if it's you," Brown said, adding that Meta's Muse agent is working on self-identification. Brown confirmed Home Depot is in active conversations with Meta about how Muse should introduce itself to websites, and said there is openness on both sides because Meta wants the agent to grow.
Several Silicon Valley companies have launched consumer-focused AI agents, including Meta's Muse and Spear Street Technology's Instinct AI, which promise to scan the internet, compare products, fill carts and buy items with little intervention. Retailers have responded differently: Amazon barred Muse from its marketplace, saying the bots did not properly identify themselves, while Walmart, Best Buy, Sephora and Shopify have announced integrations or openness to AI agents. In October, Meta said it co-developed a personal agent protocol with Sierra AI to standardize agent identification.
Brown said Home Depot does not want to lose its connection with loyal customers, including contractors and remodelers enrolled in its Pro Rewards platform, where they are eligible for discounts.
On cybersecurity, Brown said Home Depot used AI during a wave of patching last week after critical vulnerabilities were found in PHP, a programming language used in an estimated 70% of websites. The flaws were discovered with help from OpenAI and prompted US cybersecurity officials to urge organizations to update systems or take them offline. Brown called the hours after the warning "a little hectic" but said the main website was not affected. A few smaller operating units had to plug vulnerabilities, including an internal job application system taken down for about an hour to patch, she said.
Brown confirmed Home Depot has access to Anthropic's Mythos AI model but was not granted early access through Project Glasswing, whose slots were largely reserved for banks and cybersecurity firms. She said the Anthropic models found a few things to fix in Home Depot's code as well as some false positives, and that AI is pushing the company to cut the time it takes to deploy a fix to hours.