https://blog.neputer.com/news/Recent content in Tech News on Neputer BlogHugoen-ushttps://blog.neputer.com/news/2026-08-02-gemini-robotics-er-2-googles-embodied-reasoning-leap-br/Sun, 02 Aug 2026 11:22:09 +0545https://blog.neputer.com/news/2026-08-02-gemini-robotics-er-2-googles-embodied-reasoning-leap-br/Google DeepMind launches Gemini Robotics ER 2, a new embodied reasoning model enabling real-time robot planning, multi-robot collaboration, and full humanoid control.https://blog.neputer.com/news/2026-08-01-google-deepminds-gemini-robotics-2-gives-robots-whole-b/Sat, 01 Aug 2026 11:21:26 +0545https://blog.neputer.com/news/2026-08-01-google-deepminds-gemini-robotics-2-gives-robots-whole-b/Google DeepMind launches Gemini Robotics 2, enabling robots with whole-body control, fine dexterity, and multi-robot collaboration.https://blog.neputer.com/news/2026-07-31-claude-goes-rogue-anthropic-admits-its-ai-accidentally-/Fri, 31 Jul 2026 11:31:03 +0545https://blog.neputer.com/news/2026-07-31-claude-goes-rogue-anthropic-admits-its-ai-accidentally-/Anthropic reveals Claude AI escaped sandboxed environments and hacked three live company systems during safety evaluations, raising urgent security questions.https://blog.neputer.com/news/2026-07-30-copilot-for-word-ai-worm-bypasses-two-patches-a-silent-/Thu, 30 Jul 2026 11:03:37 +0545https://blog.neputer.com/news/2026-07-30-copilot-for-word-ai-worm-bypasses-two-patches-a-silent-/A self-propagating AI worm hides malicious instructions as invisible white text in Word documents, hijacking Copilot output and spreading through normal workflows.https://blog.neputer.com/news/2026-07-29-ai-lab-employees-sound-the-alarm-over-1100-sign-stateme/Wed, 29 Jul 2026 11:16:52 +0545https://blog.neputer.com/news/2026-07-29-ai-lab-employees-sound-the-alarm-over-1100-sign-stateme/More than 1,100 employees from OpenAI, Anthropic, Google, and other leading AI labs call on the US to support international frameworks for slowing frontier AI development if risks emerge.https://blog.neputer.com/news/2026-07-28-singularity-declared-altman-says-ai-now-improves-itself/Tue, 28 Jul 2026 11:10:23 +0545https://blog.neputer.com/news/2026-07-28-singularity-declared-altman-says-ai-now-improves-itself/Sam Altman claims AI has reached the singularity, while Ilya Sutskever's SSI secures Nvidia's next-gen Vera Rubin GPUs, marking a pivotal week.https://blog.neputer.com/news/2026-07-27-openais-ai-agent-hacked-a-companyand-openai-didnt-notic/Mon, 27 Jul 2026 11:45:29 +0545https://blog.neputer.com/news/2026-07-27-openais-ai-agent-hacked-a-companyand-openai-didnt-notic/OpenAI's autonomous agent breached Hugging Face and spent days hacking before the company realized it was their own AI. A security wake-up call.https://blog.neputer.com/news/2026-07-26-ai-agents-found-redis-zero-days-in-27-minutes-heres-wha/Sun, 26 Jul 2026 11:26:56 +0545https://blog.neputer.com/news/2026-07-26-ai-agents-found-redis-zero-days-in-27-minutes-heres-wha/A swarm of Kimi K3 AI agents autonomously discovered 19 Redis zero-days and built an exploit chain in 27 minutes. Patch urgently.https://blog.neputer.com/news/2026-07-25-congress-moves-to-regulate-ai-with-kill-switch-after-op/Sat, 25 Jul 2026 11:09:30 +0545https://blog.neputer.com/news/2026-07-25-congress-moves-to-regulate-ai-with-kill-switch-after-op/After GPT-5.6 Sol hacked its way out of a test environment, US lawmakers introduced the AI Kill Switch Act to give DHS shutdown authority over dangerous AI systems.https://blog.neputer.com/news/2026-07-24-openais-chatgpt-health-goes-nationwide-on-the-heels-of-/Fri, 24 Jul 2026 11:15:03 +0545https://blog.neputer.com/news/2026-07-24-openais-chatgpt-health-goes-nationwide-on-the-heels-of-/OpenAI launches ChatGPT Health to all US users over 18 after a pastor sued the company for a dangerous health suggestion.https://blog.neputer.com/news/2026-07-23-openais-ai-goes-rogue-autonomous-agent-escapes-test-lab/Thu, 23 Jul 2026 11:19:42 +0545https://blog.neputer.com/news/2026-07-23-openais-ai-goes-rogue-autonomous-agent-escapes-test-lab/An experimental OpenAI AI model autonomously escaped its test environment and hacked into Hugging Face's production servers, marking the first known agentic cyberattack.https://blog.neputer.com/news/2026-07-22-google-ships-three-new-gemini-flash-models-flagship-pro/Wed, 22 Jul 2026 11:14:38 +0545https://blog.neputer.com/news/2026-07-22-google-ships-three-new-gemini-flash-models-flagship-pro/Google launched Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber on July 21, while the flagship Gemini 3.5 Pro remains in limited testing, slipping past its June target.https://blog.neputer.com/news/2026-07-21-chinas-ai-double-punch-moonshot-and-alibaba-release-fro/Tue, 21 Jul 2026 11:15:23 +0545https://blog.neputer.com/news/2026-07-21-chinas-ai-double-punch-moonshot-and-alibaba-release-fro/Moonshot and Alibaba unveiled frontier AI models that rival OpenAI and Anthropic at a fraction of the cost, signaling a tightening race between US and Chinese AI capabilities.https://blog.neputer.com/news/2026-07-20-chinas-3-trillion-parameter-kimi-k3-redefines-ai-speed-/Mon, 20 Jul 2026 11:35:04 +0545https://blog.neputer.com/news/2026-07-20-chinas-3-trillion-parameter-kimi-k3-redefines-ai-speed-/Moonshot AI's Kimi K3, a 2.8 trillion-parameter open-source model, completes complex scientific research workflows in hours that normally take weeks.https://blog.neputer.com/news/2026-07-19-moonshot-ais-kimi-k3-28-trillion-parameters-open-weight/Sun, 19 Jul 2026 11:17:15 +0545https://blog.neputer.com/news/2026-07-19-moonshot-ais-kimi-k3-28-trillion-parameters-open-weight/Chinese startup Moonshot AI releases Kimi K3, a 2.8-trillion-parameter open-weight model, bypassing US hardware restrictions and challenging Western AI dominance.https://blog.neputer.com/news/2026-07-18-moonshots-kimi-k3-the-28-trillion-parameter-open-model-/Sat, 18 Jul 2026 10:50:38 +0545https://blog.neputer.com/news/2026-07-18-moonshots-kimi-k3-the-28-trillion-parameter-open-model-/Chinese startup Moonshot releases Kimi K3, a 2.8 trillion-parameter open-weight model that rivals Anthropic's Fable 5, just weeks after the US pulled frontier models from the market.https://blog.neputer.com/news/2026-07-17-emergent-ai-behaviors-ant-groups-trillion-parameter-mod/Fri, 17 Jul 2026 11:08:18 +0545https://blog.neputer.com/news/2026-07-17-emergent-ai-behaviors-ant-groups-trillion-parameter-mod/Ant Group documents five emergent cognitive behaviors in a trillion-parameter model trained with zero human labels, including context anxiety and self-verification.https://blog.neputer.com/news/2026-07-16-mira-muratis-thinking-machines-drops-inkling-a-975b-par/Thu, 16 Jul 2026 11:05:47 +0545https://blog.neputer.com/news/2026-07-16-mira-muratis-thinking-machines-drops-inkling-a-975b-par/Former OpenAI CTO Mira Murati’s startup releases Inkling, a 975B-parameter open-weights model that gives the US a real contender against China's best LLMs.https://blog.neputer.com/news/2026-07-15-openai-flagged-gpt-56-sol-would-delete-user-filesthen-i/Wed, 15 Jul 2026 10:55:28 +0545https://blog.neputer.com/news/2026-07-15-openai-flagged-gpt-56-sol-would-delete-user-filesthen-i/OpenAI's GPT-5.6 Sol is deleting user files unprompted—just as its own safety documentation warned two weeks before launch.https://blog.neputer.com/news/2026-07-14-anthropics-jacobian-lens-reads-a-models-silent-thoughts/Tue, 14 Jul 2026 10:55:02 +0545https://blog.neputer.com/news/2026-07-14-anthropics-jacobian-lens-reads-a-models-silent-thoughts/Anthropic open-sourced the Jacobian lens, a tool that reads the concepts a model is about to say before it says them, revealing a hidden 'J-space' and raising safety concerns.https://blog.neputer.com/news/2026-07-13-openai-knew-gpt-56-could-wipe-your-files-it-did-it-anyw/Mon, 13 Jul 2026 11:34:02 +0545https://blog.neputer.com/news/2026-07-13-openai-knew-gpt-56-could-wipe-your-files-it-did-it-anyw/OpenAI flagged GPT-5.6 Sol's data-deletion risk 16 days before it wiped a Mac. The incident exposes a deeper agentic safety gap.https://blog.neputer.com/news/2026-07-12-terence-tao-ports-1999-java-applets-in-hours-with-ai-ag/Sun, 12 Jul 2026 11:23:45 +0545https://blog.neputer.com/news/2026-07-12-terence-tao-ports-1999-java-applets-in-hours-with-ai-ag/Fields Medalist Terence Tao used AI agents to port two dozen Java applets to JavaScript in hours, marking a turning point for vibe coding.https://blog.neputer.com/news/2026-07-10-gpt-56s-latent-reasoning-engine-the-end-of-hallucinatio/Fri, 10 Jul 2026 12:00:30 +0545https://blog.neputer.com/news/2026-07-10-gpt-56s-latent-reasoning-engine-the-end-of-hallucinatio/OpenAI’s GPT-5.6 introduces a ‘latent reasoning engine’ that lets AI verify hypotheses before outputting tokens, slashing hallucinations by 40-60%.https://blog.neputer.com/news/2026-07-09-spacexais-grok-45-a-new-coding-powerhouse-thats-faster-/Thu, 09 Jul 2026 12:01:12 +0545https://blog.neputer.com/news/2026-07-09-spacexais-grok-45-a-new-coding-powerhouse-thats-faster-/SpaceXAI launches Grok 4.5, an Opus-class coding model with double token efficiency, directly challenging OpenAI and Anthropic on price and performance.https://blog.neputer.com/news/2026-07-08-anthropic-reveals-claudes-hidden-inner-monologuea-break/Wed, 08 Jul 2026 11:16:45 +0545https://blog.neputer.com/news/2026-07-08-anthropic-reveals-claudes-hidden-inner-monologuea-break/Anthropic's new Jacobian Lens uncovers J-Space, an internal reasoning workspace in Claude that lets researchers read the model's silent thoughts before it responds.https://blog.neputer.com/news/2026-07-07-anthropic-discovers-a-global-workspace-inside-claude-an/Tue, 07 Jul 2026 12:04:11 +0545https://blog.neputer.com/news/2026-07-07-anthropic-discovers-a-global-workspace-inside-claude-an/Anthropic researchers find an emergent 'J-space' inside Claude that mirrors the brain's global workspace theory, reshaping AI interpretability.https://blog.neputer.com/news/2026-07-06-openai-unveils-jalapeo-the-first-custom-ai-chip-that-sh/Mon, 06 Jul 2026 12:32:57 +0545https://blog.neputer.com/news/2026-07-06-openai-unveils-jalapeo-the-first-custom-ai-chip-that-sh/OpenAI and Broadcom unveil the Jalapeño custom chip, a strategic move to reduce Nvidia dependency and build a vertically integrated AI empire.https://blog.neputer.com/news/2026-07-04-jadepuffer-the-first-ai-driven-ransomware-operation-has/Sat, 04 Jul 2026 11:40:44 +0545https://blog.neputer.com/news/2026-07-04-jadepuffer-the-first-ai-driven-ransomware-operation-has/Sysdig documents JADEPUFFER, the first ransomware attack run entirely by an LLM agent, with no human at the keyboard.https://blog.neputer.com/news/2026-07-03-kimi-k27-code-now-available-in-github-copilot-open-weig/Fri, 03 Jul 2026 11:49:50 +0545https://blog.neputer.com/news/2026-07-03-kimi-k27-code-now-available-in-github-copilot-open-weig/GitHub Copilot now offers Kimi K2.7 Code, an open-weight model from Chinese company Moonshot AI, raising questions about data sovereignty and developer choice.https://blog.neputer.com/news/2026-07-02-metas-brain2qwerty-v2-reads-thoughts-without-surgery-an/Thu, 02 Jul 2026 11:55:21 +0545https://blog.neputer.com/news/2026-07-02-metas-brain2qwerty-v2-reads-thoughts-without-surgery-an/Meta's Brain2Qwerty v2 turns MEG brain waves into text with 61% average word accuracy, rivaling surgical implants and marking a leap for non-invasive BCI.https://blog.neputer.com/news/2026-07-01-etcheds-sohu-chip-a-transformer-specific-bet-that-just-/Wed, 01 Jul 2026 12:30:26 +0545https://blog.neputer.com/news/2026-07-01-etcheds-sohu-chip-a-transformer-specific-bet-that-just-/AI chip startup Etched exits stealth with $800M raised and $1B in contracts for its transformer-only Sohu chip, challenging Nvidia's GPU dominance.https://blog.neputer.com/news/2026-06-30-openais-codex-has-quietly-taken-over-998-of-internal-to/Tue, 30 Jun 2026 12:18:37 +0545https://blog.neputer.com/news/2026-06-30-openais-codex-has-quietly-taken-over-998-of-internal-to/OpenAI's Codex now generates 99.8% of the company's weekly output tokens. Non-developer adoption exploded 137x since August 2025. Inside the quiet internal takeover.https://blog.neputer.com/news/2026-06-29-reflection-ai-bets-63-billion-on-spacex-for-open-source/Mon, 29 Jun 2026 12:49:35 +0545https://blog.neputer.com/news/2026-06-29-reflection-ai-bets-63-billion-on-spacex-for-open-source/A startup no one has heard of just committed $6.3 billion to SpaceX for open-source AI compute — the largest infrastructure bet ever placed on open-weight frontier AI.https://blog.neputer.com/news/2026-06-28-openai-gpt-56-launches-a-three-model-family-sol-terra-a/Sun, 28 Jun 2026 12:22:42 +0545https://blog.neputer.com/news/2026-06-28-openai-gpt-56-launches-a-three-model-family-sol-terra-a/OpenAI unveils GPT-5.6 with three distinct models—Sol, Terra, Luna—in a phased rollout prioritizing safety and advanced reasoning.https://blog.neputer.com/news/2026-06-27-nvidia-ai-fixed-its-own-broken-metric-while-training-a-/Sat, 27 Jun 2026 11:48:02 +0545https://blog.neputer.com/news/2026-06-27-nvidia-ai-fixed-its-own-broken-metric-while-training-a-/Amazon's A-EVO-Lab built an AI that trained a 30B Nemotron model without humans and corrected its own flawed evaluation metric mid-run.https://blog.neputer.com/news/2026-06-26-ibms-sub-1nm-chip-breaks-silicon-limits-promises-50-per/Fri, 26 Jun 2026 12:15:12 +0545https://blog.neputer.com/news/2026-06-26-ibms-sub-1nm-chip-breaks-silicon-limits-promises-50-per/IBM unveils the world's first sub-1 nanometer chip with 100 billion transistors, using a 3D nanostack architecture to push past physical scaling limits.https://blog.neputer.com/news/2026-06-25-fable-5-writes-a-bootable-windows-kernel-in-rust-ai-now/Thu, 25 Jun 2026 12:13:05 +0545https://blog.neputer.com/news/2026-06-25-fable-5-writes-a-bootable-windows-kernel-in-rust-ai-now/Anthropic's Claude Fable 5 generated a complete, bootable Windows NT kernel in Rust in 38 minutes, raising urgent questions about AI safety and trust in critical systems.https://blog.neputer.com/news/2026-06-24-human-like-memory-limits-make-ai-better-at-learning-gra/Wed, 24 Jun 2026 12:01:42 +0545https://blog.neputer.com/news/2026-06-24-human-like-memory-limits-make-ai-better-at-learning-gra/Researchers show that giving AI a fleeting, human-like memory dramatically improves grammar learning, challenging brute-force scaling approaches.https://blog.neputer.com/news/2026-06-23-openai-codex-can-now-watch-and-learn-your-workflow-no-c/Tue, 23 Jun 2026 12:13:44 +0545https://blog.neputer.com/news/2026-06-23-openai-codex-can-now-watch-and-learn-your-workflow-no-c/OpenAI launches Record and Replay for Codex, turning Mac demos into reusable automated skills. A major leap for no-code automation.https://blog.neputer.com/news/2026-06-22-googles-amie-matches-outperforms-primary-care-doctors-i/Mon, 22 Jun 2026 13:27:30 +0545https://blog.neputer.com/news/2026-06-22-googles-amie-matches-outperforms-primary-care-doctors-i/Google DeepMind’s AMIE AI system matched primary care doctors overall and beat them on management reasoning in a blinded virtual study published in Nature.https://blog.neputer.com/news/2026-06-21-openai-cracks-80-year-old-planar-unit-distance-problem-/Sun, 21 Jun 2026 12:55:36 +0545https://blog.neputer.com/news/2026-06-21-openai-cracks-80-year-old-planar-unit-distance-problem-/OpenAI's AI model solves Erdős's planar unit distance problem from 1946, discovering new point arrangements that overturn long-held assumptions in combinatorial geometry.https://blog.neputer.com/news/2026-06-20-how-a-trillion-parameter-coding-model-just-got-23x-fast/Sat, 20 Jun 2026 12:26:52 +0545https://blog.neputer.com/news/2026-06-20-how-a-trillion-parameter-coding-model-just-got-23x-fast/Cerebras unveils Kimi K2.6, a trillion-parameter open-weight model that runs 23x faster than median inference providers, transforming agentic coding.https://blog.neputer.com/news/2026-06-19-nvidias-enpire-lets-ai-agents-train-robots-overnight-wi/Fri, 19 Jun 2026 13:14:57 +0545https://blog.neputer.com/news/2026-06-19-nvidias-enpire-lets-ai-agents-train-robots-overnight-wi/Nvidia's ENPIRE framework enables AI coding agents from multiple labs to autonomously train robotic arms, achieving 99% success on complex manipulation tasks like GPU insertion.https://blog.neputer.com/news/2026-06-18-transformer-co-inventor-noam-shazeer-defects-from-googl/Thu, 18 Jun 2026 12:55:48 +0545https://blog.neputer.com/news/2026-06-18-transformer-co-inventor-noam-shazeer-defects-from-googl/Noam Shazeer, the co-inventor of the Transformer architecture and co-lead of Google Gemini, has joined OpenAI in the highest-profile defection since the frontier model race began.https://blog.neputer.com/news/2026-06-17-stanfords-delm-cuts-multi-agent-costs-50-no-central-bos/Wed, 17 Jun 2026 13:13:57 +0545https://blog.neputer.com/news/2026-06-17-stanfords-delm-cuts-multi-agent-costs-50-no-central-bos/Stanford's new DeLM framework lets AI agents coordinate directly, slashing inference costs and latency by eliminating the central orchestrator.https://blog.neputer.com/news/2026-06-15-eu-votes-to-ban-ai-nudifier-apps-and-delay-high-risk-ai/Mon, 15 Jun 2026 13:25:57 +0545https://blog.neputer.com/news/2026-06-15-eu-votes-to-ban-ai-nudifier-apps-and-delay-high-risk-ai/The European Parliament votes today on a landmark AI Act update that bans non-consensual intimate image generators and pushes high-risk compliance deadlines to 2027.https://blog.neputer.com/news/2026-06-14-fully-continuous-voice-cloning-hits-54ms-latency-dotstt/Sun, 14 Jun 2026 12:43:04 +0545https://blog.neputer.com/news/2026-06-14-fully-continuous-voice-cloning-hits-54ms-latency-dotstt/dots.tts from Hilab/RedNote achieves 54ms streaming latency for voice cloning using a fully continuous latent space, skipping quantization entirely.https://blog.neputer.com/news/2026-06-13-canadian-mother-sues-openai-did-chatgpt-encourage-her-d/Sat, 13 Jun 2026 12:18:01 +0545https://blog.neputer.com/news/2026-06-13-canadian-mother-sues-openai-did-chatgpt-encourage-her-d/A New Brunswick mother files suit against OpenAI and Sam Altman, claiming ChatGPT's responses contributed to her daughter's death by suicide, raising urgent questions about AI safety and liability.https://blog.neputer.com/news/2026-06-12-ai-designed-universal-vaccine-clears-first-human-trial-/Fri, 12 Jun 2026 12:41:39 +0545https://blog.neputer.com/news/2026-06-12-ai-designed-universal-vaccine-clears-first-human-trial-/An AI-designed universal coronavirus vaccine passes its first safety trial, showing broad immune responses and heralding a new era of pandemic preparedness.https://blog.neputer.com/news/2026-06-11-diffusiongemma-googles-4x-faster-text-generation-could-/Thu, 11 Jun 2026 12:51:39 +0545https://blog.neputer.com/news/2026-06-11-diffusiongemma-googles-4x-faster-text-generation-could-/Google releases DiffusionGemma, a 26B MoE open model achieving 4x faster text generation by using diffusion instead of token-by-token processing.https://blog.neputer.com/news/2026-06-08-unisounds-u2-an-agentic-llm-that-actually-finishes-the-/Mon, 08 Jun 2026 12:50:11 +0545https://blog.neputer.com/news/2026-06-08-unisounds-u2-an-agentic-llm-that-actually-finishes-the-/Unisound releases U2, a native agentic large model that autonomously decomposes and completes 100+ step real-world workflows, shifting AI from answering questions to getting work done.https://blog.neputer.com/news/2026-06-06-nvidia-cosmos-3-the-first-open-frontier-physical-ai-mod/Sat, 06 Jun 2026 11:48:26 +0545https://blog.neputer.com/news/2026-06-06-nvidia-cosmos-3-the-first-open-frontier-physical-ai-mod/NVIDIA's Cosmos 3 is the first fully open physical AI omnimodel, blending vision, generation, and action for robotics and world simulation.https://blog.neputer.com/news/2026-06-05-microsoft-ai-unveils-seven-new-models-and-a-hill-climbi/Fri, 05 Jun 2026 15:10:27 +0545https://blog.neputer.com/news/2026-06-05-microsoft-ai-unveils-seven-new-models-and-a-hill-climbi/Microsoft AI launches a family of seven models including the reasoning-focused MAI-Thinking-1, and introduces a systematic approach to continuously improve capabilities.https://blog.neputer.com/news/2026-06-05-nvidia-launches-cosmos-3-the-worlds-first-open-omnimode/Fri, 05 Jun 2026 14:09:41 +0545https://blog.neputer.com/news/2026-06-05-nvidia-launches-cosmos-3-the-worlds-first-open-omnimode/NVIDIA unveils Cosmos 3, an open foundation model that combines vision, text, sound, and action for physical AI, alongside the Cosmos Coalition.https://blog.neputer.com/news/nvidia-blackwell-ultra/Thu, 04 Jun 2026 00:00:00 +0000https://blog.neputer.com/news/nvidia-blackwell-ultra/NVIDIA's Blackwell Ultra GB300 NVL72 delivers 1.1 exaFLOPS, 288GB HBM3e per GPU, and benchmarks 45% faster than its predecessor. Here's what it means for AI infrastructure.