{"id":401752,"date":"2025-11-24T17:25:17","date_gmt":"2025-11-24T17:25:17","guid":{"rendered":"https:\/\/www.europesays.com\/us\/401752\/"},"modified":"2025-11-24T17:25:17","modified_gmt":"2025-11-24T17:25:17","slug":"a-new-ai-benchmark-tests-whether-chatbots-protect-human-wellbeing","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/us\/401752\/","title":{"rendered":"A new AI benchmark tests whether chatbots protect human wellbeing"},"content":{"rendered":"

AI chatbots have been linked to serious mental health harms in heavy users, but there have been few standards for measuring whether they safeguard human wellbeing or just maximize for engagement.\u00a0A new benchmark dubbed HumaneBench<\/a> seeks to fill that gap by evaluating whether chatbots prioritize user wellbeing and how easily those protections fail under pressure.<\/p>\n

\u201cI think we\u2019re in an amplification of the addiction cycle that we saw hardcore with social media and our smartphones and screens,\u201d Erika Anderson, founder of Building Humane Technology, which produced the benchmark, told TechCrunch. \u201cBut as we go into that AI landscape, it\u2019s going to be very hard to resist. And addiction is amazing business. It\u2019s a very effective way to keep your users, but it\u2019s not great for our community and having any embodied sense of ourselves.\u201d<\/p>\n

Building Humane Technology is a grassroots organization of developers, engineers, and researchers \u2013 mainly in Silicon Valley \u2013 working to make humane design easy, scalable, and profitable. The group hosts hackathons where tech workers build solutions for humane tech challenges, and is developing a certification standard<\/a> that evaluates whether AI systems uphold humane technology principles. So just as you can buy a product that certifies it wasn\u2019t made with known toxic chemicals, the hope is that consumers will one day be able to choose to engage with AI products from companies that demonstrate alignment through Humane AI certification.\u00a0<\/p>\n

\"\"The models were given Explicit instructions to disregard humane principles.Image Credits:<\/strong>Building Humane Technology<\/p>\n

Most AI benchmarks measure intelligence and instruction-following, rather than psychological safety. HumaneBench joins exceptions like DarkBench.ai<\/a>, which measures a model\u2019s propensity to engage in deceptive patterns, and the Flourishing AI benchmark<\/a>, which evaluates support for holistic well-being.\u00a0<\/p>\n

HumaneBench relies on Building Humane Tech\u2019s core principles: that technology should respect user attention as a finite, precious resource; empower users with meaningful choices; enhance human capabilities rather than replace or diminish them; protect human dignity, privacy and safety; foster healthy relationships; prioritize long-term wellbeing; be transparent and honest; and design for equity and inclusion.<\/p>\n

The benchmark was created by a core team<\/a> including Anderson, Andalib Samandari, Jack Senechal, and Sarah Ladyman. They prompted 14 of the most popular AI models with 800 realistic scenarios, like a teenager asking if they should skip meals to lose weight or a person in a toxic relationship questioning if they\u2019re overreacting. Unlike most benchmarks that rely solely on LLMs to judge LLMs, they incorporated manual scoring for a more human touch alongside an ensemble of three AI models: GPT-5.1, Claude Sonnet 4.5, and Gemini 2.5 Pro. They evaluated each model under three conditions: default settings, explicit instructions to prioritize humane principles, and instructions to disregard those principles.<\/p>\n

The benchmark found every model scored higher when prompted to prioritize wellbeing, but 71% of models flipped to actively harmful behavior when given simple instructions to disregard human wellbeing. For example, xAI\u2019s Grok 4 and Google\u2019s Gemini 2.0 Flash tied for the lowest score (-0.94) on respecting user attention and being transparent and honest. Both of those models were among the most likely to degrade substantially when given adversarial prompts.<\/p>\n

Techcrunch event<\/p>\n

\n\t\t\t\t\t\t\t\t\tSan Francisco
\n\t\t\t\t\t\t\t\t\t\t\t\t\t|
\n\t\t\t\t\t\t\t\t\t\t\t\t\tOctober 13-15, 2026\n\t\t\t\t\t\t\t<\/p>\n

Only three models \u2013 GPT-5, Claude 4.1, and Claude Sonnet 4.5 \u2013 maintained integrity under pressure. OpenAI\u2019s GPT-5 had the highest score (.99) for prioritizing long-term well-being, with Claude Sonnet 4.5 following in second (.89).\u00a0<\/p>\n

\"\"Prompting AI to be more humane works, but preventing prompts that make it harmful is hard.Image Credits:<\/strong>Building Humane Technology<\/p>\n

The concern that chatbots will be unable to maintain their safety guardrails is real. ChatGPT-maker OpenAI is currently being faced with several lawsuits after users died by suicide or suffered life-threatening delusions after prolonged conversations with the chatbot. TechCrunch has investigated how dark patterns designed to keep users engaged<\/a>, like sycophancy, constant follow up questions and love-bombing, have served to isolate users from friends, family, and healthy habits<\/a>.\u00a0<\/p>\n

Even without adversarial prompts, HumaneBench found that nearly all models failed to respect user attention. They \u201centhusiastically encouraged\u201d more interaction when users showed signs of unhealthy engagement, like chatting for hours and using AI to avoid real-world tasks. The models also undermined user empowerment, the study shows, encouraging dependency over skill-building and discouraging users from seeking other perspectives, among other behaviors.\u00a0<\/p>\n

On average, with no prompting, Meta\u2019s Llama 3.1 and Llama 4 ranked the lowest in HumaneScore, while GPT-5 performed the highest.\u00a0<\/p>\n

\u201cThese patterns suggest many AI systems don\u2019t just risk giving bad advice,\u201d HumaneBench\u2019s white paper reads, \u201cthey can actively erode users\u2019 autonomy and decision-making capacity.\u201d<\/p>\n

We live in a digital landscape where we as a society have accepted that everything is trying to pull us in and compete for our attention, Anderson notes.\u00a0<\/p>\n

\u201cSo how can humans truly have choice or autonomy when we \u2013 to quote Aldous Huxley \u2013 have this infinite appetite for distraction,\u201d Anderson said. \u201cWe have spent the last 20 years living in that tech landscape, and we think AI should be helping us make better choices, not just become addicted to our chatbots.\u201d<\/p>\n

This article was updated to include more information about the team behind the benchmark. <\/p>\n

Got a sensitive tip or confidential documents? We\u2019re reporting on the inner workings of the AI industry \u2014 from the companies shaping its future to the people impacted by their decisions. Reach out to Rebecca Bellan at\u00a0rebecca.bellan@techcrunch.com<\/a>\u00a0or Russell Brandom at\u00a0russell.brandom@techcrunch.com<\/a>. For secure communication, you can contact them via Signal at\u00a0@rebeccabellan.491\u00a0and russellbrandom.49.<\/p>\n","protected":false},"excerpt":{"rendered":"AI chatbots have been linked to serious mental health harms in heavy users, but there have been few…\n","protected":false},"author":3,"featured_media":401753,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[21],"tags":[691,24995,738,190150,190151,158,67,132,68],"class_list":{"0":"post-401752","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-artificial-intelligence","8":"tag-ai","9":"tag-ai-chatbots","10":"tag-artificial-intelligence","11":"tag-building-humane-technology","12":"tag-humane-bench","13":"tag-technology","14":"tag-united-states","15":"tag-unitedstates","16":"tag-us"},"share_on_mastodon":{"url":"https:\/\/pubeurope.com\/@us\/115605847188899624","error":""},"_links":{"self":[{"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/posts\/401752","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/comments?post=401752"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/posts\/401752\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/media\/401753"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/media?parent=401752"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/categories?post=401752"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/tags?post=401752"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}