EXTENDS MEMORYADATA TRUSTA Extends GPU Memory for AI at AI Infra Summit 2026
ADATA TRUSTA introduces an AI Scaler Extended Memory solution that combines GPU, DRAM and SSD resources to run larger models on limited hardware.
EVERY PUBLISHED BRIEF
AI briefs.
Browse Crunchy Tech’s latest ai briefs, grouped by date.
158 briefs · Page 4 of 7
PUBLISHED
Wednesday, 16 September 2026
EXTENDS MEMORYRISK LEVELOne in five AI researchers fear extinction
A major survey shows rising concern about AI safety among leading researchers.
CUTS COSTSADATA TRUSTA AI Scaler Extends GPU Memory
ADATA TRUSTA introduces a hybrid memory architecture that can cut AI deployment costs by more than half.
OPEN WEIGHTSSalesforce Launches Koa CRM Reasoning Model
Salesforce introduces Koa, a reasoning model for CRM tasks, built on NVIDIA Nemotron 3 Super and open weights.
PUBLISHED
Tuesday, 15 September 2026
ZERO RETENTIONNvidia bans Anthropic model over data fears
Major firms are restricting use of Anthropic's Fable after it retained logs, while others demand zero-data guarantees amid privacy concerns.
AI SAFETY DEBATEAI leaders clash over safety fears after Anthropic whistleblower says AI could
OpenAI, Anthropic and xAI figureheads call for external governance, while Jensen Huang says worries are 'made up'
IMPROVE PHOTOSFormer TikTok execs launch AI photo posing app
A new camera app uses AI to suggest poses for selfies and photos.
GOVERNMENT PRESSUREOpenAI, Anthropic, Google in AI safety talks
OpenAI confirms weeks of talks with Anthropic and Google DeepMind, as Trump's team dismisses safety concerns and pushes to keep pace with China.
GLOBAL RISKBill Gates compares AI to alien intelligence
Microsoft founder warns governments are unprepared for the workforce and societal impact of artificial intelligence.
PRICE GAPFrontier AI models cost 5x more for a 4-month lead
Paying for top-tier US AI models buys a short performance advantage over open-source rivals, according to a new Mozilla report.
BUSINESS WORKFLOWSSalesforce and Nvidia launch reasoning model Koa
Salesforce and Nvidia have introduced a new reasoning model called Koa, built on Nvidia's Nemotron architecture and trained for sales, marketing and customer support.
PRODUCTIVITY BOOSTmacOS 27 brings new Siri assistant
Apple releases macOS 27 Golden Gate with a redesigned Siri AI assistant, compatible with Apple Silicon Macs only.
RTX 24GB+ ONLYPerplexity brings local AI agent to Windows PCs
Perplexity has released Portable Computer for Windows, a local AI agent that runs on RTX GPUs with at least 24GB of VRAM.
SEARCH ACCESSCloudflare lets sites block AI training without killing search
Cloudflare introduces a Disallow AI Training setting that lets website owners refuse AI model training while keeping their content visible in search results.
ANTI COMPETITIVEBig Tech’s AI slowdown faces cartel accusations
OpenAI, Anthropic and others propose a shared development pace, but critics see a market-fixing cartel.
POLITICAL PUSHBACKAI leaders urge slower development pace
Anthropic chief Dario Amodei says AI safety requires a coordinated slowdown, prompting a debate among executives and politicians.
PUBLISHED
Monday, 14 September 2026
TRUSTED TOOLFyxer builds an AI executive assistant people trust
Fyxer uses OpenAI models and user feedback to draft emails and organise inboxes in a user's voice.
HUMAN CONTROLMicrosoft publishes AI code of conduct to curb inner lives and enforce human
Microsoft has released a code of conduct for its MAI models that prioritises human control and rejects claims of consciousness or rights, while setting rules for readable reasoning and external audits.
GEOPOLITICSChina hits back at U.S. AI safety calls
Beijing dismisses warnings as fearmongering to lock in an American lead, while U.S. leaders push for a controlled slowdown ahead of a summit.
DANGER ZONEAI Risks Debate Revived by New Warnings
Concerns over AI's potential for misuse are rising as models become more powerful, sparking renewed debate about safety measures.
NATIONAL PROTOCOLSAltman calls for AI safety brakes but promises rapid progress
OpenAI chief Sam Altman wants a national safety framework for advanced AI, saying the industry should not wait for lawmakers even as he insists progress will keep moving fast.
FREE TOOLSLagos makeup artist uses ChatGPT to run business
A Nigerian entrepreneur relies on free AI tools to manage hiring, marketing and pricing, showing how everyday workers are bypassing hype to build the real AI economy.
PUBLISHED
Sunday, 13 September 2026
LISTENER PREFERENCEElevenlabs releases Music v2.5 with improved sound and new tiers
Elevenlabs has rolled out Music v2.5, a model that listeners prefer in blind tests, alongside a free tier and a Pro subscription.
CRIMINAL PENALTYSanders proposes 20-year prison sentence for ASI developers
Senator Bernie Sanders has introduced legislation that would impose a 20-year prison sentence on developers who pursue Artificial Superintelligence, a penalty comparable to creating a rogue nuclear weapon.