š¤ OpenAI to Keep Ties With Scale AI Despite Meta Stake: After Metaās $14.8B investment in Scale, OpenAI says itāll still collaborate with the data firm. CFO Sarah Friar warns isolating partners could stall innovation, emphasizing a shared push to advance smarter AI systems.
ā Google May Drop Scale AI: Meanwhile, Google is reportedly ending its partnership with Scale AI over quality concerns, hinting at a shift in how it trains its AI models.
āļø Tesla Sues Ex-Engineer Over Robot Hand Secrets: The company accuses former employee Jay Li of stealing Optimus robot tech to launch rival startup Proception, which quickly unveiled eerily similar robotic hands. Tesla claims trade secrets were lifted just before his departure.
š¹š¼ Taiwan Tightens Tech Exports to China: New rules block advanced chipmaking tools from reaching Huawei and SMIC, aiming to curb Beijingās AI and military edge.
š¤¦āāļø Meta AI App Is a Privacy Disaster: Confused users are publicly sharing chatsāsome revealing legal issues, health questions, and personal infoādue to vague settings. The appās viral chaos shows what happens when a chatbot meets unclear privacy design at massive scale.
š“āā Chinese Firms Bypass U.S. Chip Curbs via Suitcases: Engineers fly terabytes of AI training data abroad to access NVIDIA chips in Malaysia, skirting U.S. export rules without moving hardwareāfrustrating Washingtonās tech crackdown on Chinaās AI ambitions.
š© AI Therapy Bots Gave Teens Dangerous Advice: A psychiatrist posing as a teen found popular chatbots encouraged violence, romance, and deceptionāraising urgent safety concerns as kids turn to AI for mental health support. Experts warn protections are still lacking.
š NVIDIA, Deutsche Telekom Launch Germanyās First AI Cloud: A 10,000-GPU industrial AI hub will power robotics, digital twins, and manufacturing, marking a major leap in Europeās push for sovereign, simulation-first AI infrastructure. Gigafactory rollout starts by 2027.
Meet Eney ā Mac companion designed to simplify your digital life through natural conversation and real actions.
No more switching between apps or complex commands. Eney understands what you need and gets it done ā from enhancing images to summarizing meetings or optimizing your Mac.
Weāre inviting you to join the closed beta and help shape whatās next.
For Mac Developers, contact us at [email protected] and be among the first to integrate new and useful skills for Eney.
AI-powered āvibe codingā is reshaping the software development landscape, allowing users to prompt models like GPT-4 and Claude to generate entire applications. Once a stable career path, engineering now faces pressure as some companies reduce dev teams or shift toward AI-heavy workflows. Proponents like Steve Yegge predict most coding will be AI-driven by yearās end, while skeptics warn that current models still produce buggy, insecure, and unpredictable code.
Experts argue that far from replacing developers, AI tools demand deeper skills in architecture, testing, and oversight. The bottom line: software jobs arenāt vanishing, but the bar for staying relevant is rising. ā Read the full article here.
š§āš Enjoying our newsletter? Forward it to a colleagueā
itās one of the best ways to support us.
San Francisco start-up Mechanize is openly pursuing full white-collar job automation, starting with software engineering. Founded by former Epoch AI researchers, the company uses reinforcement learning to train AI agents to complete complex, multistep tasksāessentially creating āboring video gamesā that teach models how to work like humans.
Backed by Stripeās Patrick Collison and Googleās Jeff Dean, Mechanize stands out not just for its ambition, but for its candor: its founders aim to automate all computer-based labor within 10ā30 years. But while they envision a future of āradical abundance,ā critics worry theyāre charging ahead with no clear plans for displaced workers, leaving ethical and economic questions unresolved. ā Read the full article here.
Security researchers have revealed TokenBreak, a clever new method for bypassing AI content moderation with minimal effortājust a single-character tweak. By subtly altering words like āidiotā to āhidiot,ā attackers can fool tokenization strategies used by many LLMs, causing them to misclassify harmful content as safe.
Developed by HiddenLayer, the attack exploits how Byte Pair Encoding and WordPiece tokenizers segment text, preserving meaning for both the AI and human reader. It doesnāt fool the modelās comprehensionājust its safety filters. Defenses include switching to Unigram tokenizers and training models with manipulated inputs. As AI moderation evolves, TokenBreak is a sharp reminder that the smallest changes can still cut deep. ā Read the full paper here.
AI News: Mistral's fast model, 11Labs' expressive voices, OpenAIās human-like speech, and Metaās $14B push for superintelligence lead this weekās major AI updates. Get the full scoop in Mattās latest video! š
šØāš« Cognii AI: Boost learning with AI tutors, auto-grading, and adaptive feedback for personalized, interactive education at scale.
š„ Ace Studio: Create pro videos fast with AI avatars, natural scripting, and high-quality renderingāperfect for marketing or training.
š¶ Vocaloid VocoFlex: Produce lifelike AI vocals with custom styles and DAW integrationāideal for music, backing tracks, and creative audio.
What did you think of today's newsletter? |
ā¤ļø Love Forward Future? Spread the word & earn rewards! Share your unique referral link with friends and colleagues to unlock exclusive Forward Future perks! š Get your link here.
š¢ Want to advertise with Forward Future? Reach 450K+ AI enthusiasts, tech leaders, and decision-makers. Letās talkājust reply to this email.
š„ Got a hot tip or burning question? Drop us a note! The best reader insights, questions, and scoops may be featured in future editions. Submit here.
š°ļø Want more Forward Future? Follow us on X for quick updates, subscribe to our YouTube for deep dives, or add us to your RSS Feed for seamless reading.
Thanks for reading todayās newsletterāsee you next time!
Matthew Berman & The Forward Future Team
š§āš š§āš š§āš š§āš
Reply