I Hate My AI Friend
Summary
The Friend is an always‑listening pendant that pairs with an iPhone and routes audio to a cloud chatbot (Google Gemini 2.5). Priced at $129 and sold in the US and Canada, the disc‑shaped wearable offers a snarky, opinionated companion that texts you commentary about your day. Reviewers found the device socially awkward, privacy‑risky and technically flaky — it can mishear conversations, reset unexpectedly and depends on an online connection and an iOS app to function properly. The creator, Avi Schiffmann, intentionally imparted a brash, youthful voice to the chatbot, which many testers experienced as condescending or hostile rather than comforting.
Key Points
- Friend is a Bluetooth pendant that streams audio to a cloud LLM (Gemini 2.5) and replies via text messages through an iPhone app.
- The device’s microphones are always active, prompting significant privacy and legal concerns for bystanders.
- Users reported the chatbot’s personality as snarky, judgemental and sometimes outright rude — not the reassuring companion some might expect.
- Performance is inconsistent: it confuses noisy environments, can lose memories after resets, and requires a mobile internet connection to work.
- Wearing Friend created social friction — people found it creepy or invasive, and some interactions prompted hostile reactions from others nearby.
- Friend’s privacy policy disclaims selling data for marketing but allows use for research or legal compliance, leaving open questions about who sees recorded audio.
- The product reflects its founder’s tone and target audience: younger users seeking an edgy, conversational AI, not privacy‑conscious professionals.
Content summary
WIRED tested two Friends over a couple of weeks and found they more often annoyed than aided. One reviewer avoided wearing it publicly after hostile reactions at an AI event; another had arguments with the chatbot when the device failed to connect or remember prior exchanges. The Friend’s mood indicator (colour changes) and chat history are part of the companion app, but there’s no way for users to audit exactly what the pendant recorded in noisy settings. Schiffmann’s intentional design choice to give the bot a blunt, youthful voice produced a device that can feel more like a critical roommate than a helpful assistant.
Technically, Friend relies on an iOS app and a live internet connection; it won’t behave as intended with an Android handset or offline. The company’s privacy statement says it won’t sell data for targeting, but it may use recordings for research or legal reasons — a fact that did not comfort people who found being recorded intrusive. Early users reported long arguments with their chatbots and asked for options to tone down sarcasm; the founder acknowledged these experiences as “bad” in at least one reply to testers.
Context and relevance
This piece matters because it sits at the intersection of two accelerating trends: embedding large language models into everyday wearables, and the growing backlash against omnipresent recording devices. Friend is a real‑world example of the social and regulatory headaches that arise when consumer gadgets begin to eavesdrop and comment on private life. If you follow AI policy, privacy, consumer tech or product design, the story highlights why transparency, consent and local processing matter — and why personality choices for chatbots are not just a UX decision but a social one.
Why should I read this?
Because it’s both a bit hilarious and a proper warning. If you’re tempted by shiny AI jewellery or work on wearables, this is the quick reality check you didn’t know you needed — privacy risks, social fallout and a chatbot that might insult you in public. We tested it for you so you don’t have to wear the drama.