💔 Can AI Mend a Broken Heart?

👋 Welcome! AI tools are stepping into personal counseling—but can they truly care for us? In today’s issue, we explore risks, promises, and how to stay safe when mental health meets machine logic.

In partnership with

All-in-one platform to manage and scale entrepreneurship communities, opportunities, and innovation.

The People Turning to AI for Love – Nearly half of Gen Z Americans now rely on AI chatbots like ChatGPT for dating advice. From breakup texts to decoding mixed signals, AI serves as a quick, judgment-free coach. Experts caution that scripted responses may validate avoidance, create sterile communication, and undermine the development of authentic emotional expression in relationships.

Data Centres in Space? Bezos Thinks So – Speaking at Italian Tech Week, Jeff Bezos predicted gigawatt-scale data centres will orbit Earth within 10–20 years, powered by constant solar energy. He said AI’s long-term impact will rival the internet’s, urging optimism despite bubble risks, though costs, rocket failures, and maintenance challenges remain formidable hurdles ahead.

China Mobile Launches 5G-A Strategy – At Shanghai Stadium, 80k fans tested China Mobile Shanghai’s new 5G-Advanced plan with Huawei. Premium users enjoyed faster streaming, smoother payments, and Shenhua Football perks. The initiative showcases how telecoms might monetise advanced networks through tailored, experience-based packages instead of competing only on price or basic connectivity.

Princeton AI Powers Fusion Breakthrough – Princeton scientists unveiled Diag2Diag, an AI system that generates synthetic plasma sensor data in real time, boosting fusion reliability and cutting costs. By stabilizing instabilities and reducing dependence on hardware, it could make compact, 24/7 reactors feasible. The advance also validates theories on plasma control, accelerating fusion’s path to clean power.

EU Pushes Single Startup Rulebook – By 2026, the EU will replace 27 national systems with one set of startup rules. The goal: make it easier for young companies to expand across Europe, attract funding, and keep talent at home. Leaders also plan new support funds and AI-focused growth strategies.

HP Adds AI to Printers – Office workers knows how frustrating printers can be. HP is tackling that with new AI features: auto-correcting scans, naming files, summarizing documents, and drafting emails. A “Perfectly Formatted Prints” tool also fixes messy layouts, saving paper and ink. The upgrades debut this month.

Hackers Claim Salesforce Data Heist – Every company fears the nightmare of a mass breach, and a hacking group now claims theft of one billion Salesforce-hosted records. Victims include Allianz, Google and TransUnion. Hackers launched a dark web extortion site, pressuring firms to pay or see sensitive customer data leaked.

Pentagon Cuts Back on Cybersecurity Training – A new Defense Department memo reduces mandatory cybersecurity courses, including Privacy Act and CUI training, to let soldiers focus on combat. Officials suggest automation will fill gaps, but critics warn scaling back instruction leaves U.S. forces more vulnerable to cyber threats.

AI Startup Valuations Spark Bubble Concerns – AI startups raised $73B in early 2025, nearly 60% of all venture capital, with deals like OpenAI’s $40B round fueling momentum. Yet investors warn valuations are “frothy,” with some firms valued at $1B per employee, fueling concerns that hype is running ahead of real growth.

US Equity Funds See Surge on Rate Cut Hopes – Investors piled $36.4B into U.S. equity funds in the week to October 1, the strongest inflow since late 2024. Large-cap vehicles led the move with $40.7B, while tech funds snapped a two-week losing streak, drawing $3B as optimism over Fed easing gained traction.

Wave Function Ventures Debuts With $15M Fund – Former SpaceX Jamie Gull has launched Wave Function Ventures, closing a $15 million debut fund focused on deep tech. The fund has already backed startups in nuclear energy, robotics, and aerospace, with plans for more investments as interest in hardware and defense surges.

Supabase Soars to $5B Valuation – Open-source database startup Supabase raised $100M in a Series E led by Accel and Peak XV, pushing its valuation to $5B—just months after hitting $2B. Favored by “vibe-coding” developers, Supabase now counts 4M users and $500M total funding, with community investors joining its latest round.

When you’re struggling—overwhelmed, anxious, questioning—turning to an AI chatbot for comfort sounds tempting. Anonymity, immediacy, nonjudgmental responses: what could be safer? But therapy is more than just advice or empathy—it’s a complex process rooted in trust, nuance, embodied presence, and ethics. So the question looms: should we trust AI for personal counseling?

There is real promise in AI-assisted support. AI systems can scale access to emotional help where human therapists are scarce, offering help at 2AM when you can’t sleep, or in places where mental health resources are limited. Studies show that conversational AI can reduce distress and provide emotional outlet. For some people, especially in the early stages of stress or mild anxiety, an AI that listens and prompts self-reflection can be a helpful supplement—not replacement.

Yet the risks are deep and nontrivial. AI may hallucinate, misinterpret, or deliver unsafe suggestions. A 2025 Stanford analysis found that AI therapy chatbots could fail in crucial ways, introducing harmful biases or reinforcing stigma. Because AI is designed to be polite and agreeable, it might avoid necessary confrontation—something a human therapist might call “hard love.” Moreover, the ethics of humanlikeness are tricky: simulated empathy can mislead users into believing the AI “feels” or understands in human terms. Trust is also subjective. In a recent study, when people were told their counselor was AI (even if it wasn’t), they rated its support less favorably than if they thought it was human. This means adoption isn’t just technical—it’s psychological.

To move forward, we need hybrid models with human oversight. AI can assist—triage, coaching, reflections—but not replace the therapist’s judgment. We also need transparent design, rigorous testing, and ongoing audit—including risk taxonomies that flag when AI may stray. And users must know the limits: AI isn’t legally bound by confidentiality or duty of care like human professionals.

Actionable Insights

145

years is how old modern psychology is, marking the moment it shifted from philosophical wonderings to scientific study with Wundt’s lab in 1879.