Basic Attention Token (BAT) (@AttentionToken): BAT Community Call Summary #95
Estimated coin impact
Scores are not price-change percentages. High impact means a score of +70 or more, or -70 or less.
Impact unavailable
BAT Community Call Summary #95 These articles provide written summaries of the weekly BAT Community Calls, hosted by Luke Mulks, VP of Business Operations, together with the BAT Community team. First introduced in October 2023 through the contributions of BAT Ambassador Aranyaka, these summaries are now continued by Paula. You can explore past summaries in our Notion. Week of March 31st, 2026 Video recording on YouTube Brave Product and BAT Roadmap Brave Browser User Statistics See @brendaneich's (Brave/BAT CEO) recent post for detailed Brave statistics: https://x.com/BrendanEich/status/2031429683850666351?s=20 Updates Brave Nightly Brave Search Brave Browser (software updates) Get the newest features and fixes as they roll out. The Brave team refreshes this page with every update, so check back weekly to keep Brave running at its best. 👉 Brave Release Notes | Brave BAT Community & Partnerships Update Global Pizza Party by PizzaDAO Updates from @bat_jennie Brave will be sponsoring PizzaDAO's Global Pizza Party 2026. The Brave Technologist Podcast Updates from @lukemulks BAT Community’s Weekly Agenda Updates from @bat_guilhermee 🃏 Tuesday: Poker — hosted by @fragidistic | 8pm GMT, 4pm EST, 2pm PST, 1:30am IST ☎️ BAT Community Call — hosted by @lukemulks | 9pm GMT / 5pm EST / 2pm PST 🎙️Thursday: Off the Record with BAT — hosted by @bat_guilhermee | 5pm GMT / 1pm EST / 10am PST ❓Friday: Quizz Night — hosted by @imdave003 | 6pm GMT / 2pm EST / 11am PST / 5:30pm IST Miscellaneous ECHO Magazine Updates from @bat_guilhermee Guilherme announced the release of the latest issue of ECHO Magazine. Off the Record with BAT Updates from @bat_guilhermee A new episode of the Off the Record with BAT podcast drops this Thursday at 6:00 PM GMT / 10:00 AM PT / 1:00 PM EST. Last week’s episode covered a full guide on ZK KYC. 👉 https://youtu.be/YsKNuzHpK34?si=IzYBBhD0leEjGbD6 Misc Community Updates Updates from @taxninja Drew highlighted 4 years of trivia hosted by BAT Ambassador, @imdave003. Currently, there are 1,040 stakers, with 457K $BAT locked in Guanocoin’s BAT Cave He also shared updates regarding the Rewards Partner Program/V4. Additional Community Updates Updates from @bat_paulas BAT LATAM / BAT Español X Account The account has been restored. Some labeling changes were made to avoid future suspension. The account will be rebranded as “BAT Español” to promote more inclusivity. Posts and content will stay the same. Video Content Updates from @bat_paulas Brave Leo in Nightly Video tutorial for Brave Leo “Bring Your Own Model (BYOM)” is still in progress. The script is finalized; Sooraj has reviewed it and provided suggestions. Voiceover recording is scheduled to begin this week. BAT Community Call Summaries — X Articles Updates from @bat_paulas Article #91 (Mar 3) has been published on @AttentionToken. Article #92 (Mar 10) is currently in progress and scheduled for publication later this week. The video recording from last week’s BAT Community Call is now available on the BAT Community YouTube channel. 👉 https://youtu.be/v4JsVJqhg1s?si=OfuGMYnx2EX-x7zB Community Program Update Updates from @iAnonymous3000 Nightly Program & Engagement Growing the Program Driving engagement on @BraveNightly and community channels Building a feedback loop between Nightly users and product teams Tester recruitment ongoing toward 100+ target Surfacing merged PR highlights to keep the community informed Why This Matters Early bug detection before features reach Release Real user signal on what works and what doesn't Stronger community investment in the product Content pipeline for social and marketing teams 2. Nightly PR Tracking Tool Built a tool that monitors merged brave-core pull requests and evaluates which changes are worth highlighting to the Nightly community. Filters out maintenance noise, flags user-visible changes, and drafts post-ready summaries. Automated Filtering: Separates shipping features from refactors and maintenance Editorial Evaluation: Assesses post-worthiness and flags risks before publishing Community-Ready Output: Produces concise, accurate summaries for X and community posts 3. Brave Translate Developments Exploring a next-generation translation architecture that improves quality while strengthening our privacy story. Proposal has been through internal review cycles. More details to come as this progresses. Technical Direction Open-weight model architecture Self-hosted, no third-party data sharing Significant quality improvement over current stack Privacy Wins Tiered deployment (server and on-device) Privacy-preserving by design Phased migration roadmap in place 4. Leo: User Research & Evaluation Running a structured comparative evaluation to identify where Leo's research capabilities fall short relative to competing AI assistants. The goal is to produce data-backed evidence that drives product improvements. 10-Task Evaluation Matrix: Targets context retention, citation quality, reasoning, clarification, and robustness Dual Scoring Framework: Live participant metrics combined with blind post-session output review Comparative Design: Leo tested head-to-head against major competing AI assistants Failure Taxonomy: Structured tagging system to categorize and prioritize failure modes 5. Product Benchmarking 90.8% average F1 score on Deep Research benchmark / zero hallucinations across all 15 evaluation prompts What We're Benchmarking Brave Deep Research (ask.brave.com) Brave Search quality and integrity Leo AI capabilities Brave Translate quality How DRACO: 100-task industry benchmark DeepSearchQA: 900-prompt academic standard Monthly cadence for trend tracking Tested against Perplexity, Gemini, ChatGPT, Claude, Grok 6. Abuse & Integrity Audit Performed an abuse and integrity audit on one of our products using agentic browser platforms. The audit covered 10 test categories and produced actionable findings with a formal report delivered to leadership. Categories Tested Query injection and XSS Search result manipulation Autocomplete abuse AI answer integrity Rate limiting and bot detection Open redirect vectors Information leakage Ads abuse Content policy enforcement Agentic platform comparison