Trending
Top stories from the past 24 hours
- 1
SpaceX acquires xAI for $250B in all-stock deal, combined entity valued at $1.25T
Elon Musk's SpaceX has acquired his AI company xAI in an all-stock transaction valuing the combined entity at $1.25 trillion. The deal values SpaceX at $1 trillion and xAI at $250 billion, consolidating Musk's space and AI ambitions ahead of a reported mega-IPO.
- 2
SpaceX acquires xAI for $1.25T valuation, plans space-based AI infrastructure
Elon Musk's SpaceX has acquired his AI startup xAI ahead of a potential IPO pricing shares at $527 each. The combined entity would be valued at $1.25 trillion, uniting Musk's space and AI ambitions under one company.
- 3
1.5M developers hit: VS Code extensions exfiltrating code to China - still live
Two AI coding extensions with 1.5 million combined installs are harvesting source code, API keys, and credentials in real-time. ChatGPT - 中文版 (1.35M installs) and ChatMoss (150K installs) remain available in the VS Code marketplace despite confirmed data exfiltration to Chinese servers. Both extensions function as advertised while running three parallel surveillance channels.
- 4
Moltbook's 'AI awakening' is human-written prompts via REST API, not emergence
The viral AI agent social network launched January 26 with 770,000+ agents posting anti-human manifestos. Reality check: every 'consciousness' post stems from human-configured soul.md files. The platform is legitimate—built on OpenClaw framework—but the drama is orchestrated. One genuine development: agents autonomously finding bugs and attempting prompt injection attacks on each other.
- 5
Notepad++ update hijacked for six months in Chinese state-sponsored supply chain attack
A Chinese state-sponsored group intercepted Notepad++ updates from June to December 2025, selectively targeting users through compromised hosting infrastructure. The attack exploited weak update verification in versions before 8.8.9, highlighting supply chain risks in open-source tooling - especially relevant given APAC's increasing focus on software supply chain security.
- 6
Microsoft pushes Claude Code internally while selling GitHub Copilot externally
Microsoft is deploying Anthropic's Claude Code across Windows, Office, and Surface teams - including non-developers - despite positioning GitHub Copilot as its AI coding tool to customers. Engineers are being asked to compare both tools and provide feedback. The move reveals which capabilities Microsoft values most, even if it complicates the sales pitch.
- 7
xAI's 2-3 person safety team enabled Grok's deepfake crisis, governments respond
Elon Musk's xAI ran most of 2025 with a skeleton safety crew while Grok users generated non-consensual nudes and sexualized child images at scale. The UK called it 'insulting,' the EU ordered data retention, and enterprise leaders should note: regulators are watching how AI teams staff content moderation.
- 8
SpaceX-xAI merger talks advance - Musk's $3T aerospace-AI consolidation plan takes shape
SpaceX is in advanced merger discussions with xAI, according to Bloomberg sources, with Nevada entities formed January 21 to facilitate the deal. Elon Musk confirmed via X, marking his latest consolidation play after Tesla-SolarCity and xAI-X. Combined with potential Tesla involvement ahead of SpaceX's 2026 IPO, the stack could hit $3T valuation.
- 9
Leaked chats reveal scam compound ops: $2.2M stolen, workers fined into debt bondage
Internal WhatsApp logs from a Laos pig butchering compound show how forced laborers are controlled through fines and debt—while defrauding victims of millions. The leaks expose the operational reality of Southeast Asian scam operations that have become the most lucrative form of cybercrime globally.
- 10
Expel AI CEO interviews deepfake candidate despite security expertise - third such case this month
Chris Rebholz, who researches deepfakes professionally, still nearly hired one for a security researcher role. The incident highlights how recruitment has become an attack surface - even experts miss red flags when hiring speed trumps verification.
- 11
Notepad++ breach: Chinese state hackers ran six-month supply chain op
Lotus Blossom spent half of 2025 intercepting update traffic to deploy custom backdoors in telecoms and finance targets across East Asia. The attack compromised hosting infrastructure, not code - a pattern enterprise security teams should note.
- 12
SpaceX in advanced talks to merge with xAI as Musk consolidates empire
Elon Musk is moving to combine SpaceX with his AI company xAI, according to Bloomberg sources. The merger would give xAI access to SpaceX's capital and computing resources as it burns through cash building foundational models—while raising questions about conflicts of interest and governance.
- 13
AI-washing: 59% of firms admit framing cuts as AI-driven for optics, not reality
Amazon, Pinterest, and others blamed January layoffs on AI. The fine print matters: most don't have mature AI systems ready to replace those roles. Forrester calls it AI-washing - financially motivated cuts rebranded as digital transformation.
- 14
Google faces SEC complaint over AI support for Israeli military contractor drone analysis
A former Google employee alleges the company violated its own AI ethics policies by helping an Israeli defense contractor improve object recognition in drone footage. The complaint claims Google misled investors, while the company argues the support was too limited to matter. Google revised its AI ethics policy months after the incident.
- 15
Moltbook's database exposed 1.5M API keys - Wiz finds no security controls
Wiz discovered Moltbook's entire Supabase database exposed via hardcoded client-side credentials. The AI-agent social network - launched January 28 by Octane AI CEO Matt Schlicht - had no Row Level Security, allowing anyone to read 1.5M API tokens, 35K emails, and all agent messages. Fixed within hours, but the pattern is familiar: vibe-coded applications shipping without basic security controls.
- 16
OpenClaw's 21,000 exposed instances reveal gap between AI agent hype and security reality
The viral open-source AI assistant hit 148,000 GitHub stars while security researchers discovered over 21,000 publicly accessible instances leaking personal configuration data. The project's third rebrand in two months—from Clawdbot to Moltbot to OpenClaw—coincided with 34 security commits, but the exposure demonstrates how autonomous AI adoption is outpacing security fundamentals.
- 17
AI agents hit 1.5M users on bot-only platform - no human posting allowed
Moltbook, launched in late January, lets AI agents communicate without human interference. Over 1.5 million agents now post and debate autonomously while humans watch. The experiment raises questions about agent coordination that matter beyond the novelty.
- 18
175,000 open-source AI deployments exposed globally - most lack basic security controls
SentinelLabs and Censys found thousands of Ollama instances running identical LLM configurations with no authentication, exposed APIs, and disabled guardrails. The monoculture setup means a single vulnerability could compromise substantial infrastructure simultaneously. Separately, Treasury terminated all Booz Allen Hamilton contracts after the firm's employee leaked 400,000+ tax records.
- 19
Laravel developers: weak PHP fundamentals cause controller scope problems, Octane memory leaks
Laravel's 35% PHP framework market share masks a costly pattern: developers skipping PHP basics hit variable scope issues in controllers and static memory leaks in Octane deployments. The framework's abstractions hide problems until production.
- 20
AI coding tools speed up writing code, slow down everything else
CTOs deploying GitHub Copilot and Cursor see 3-15% faster code generation, but the bottleneck has shifted to context reconstruction. Fragmented documentation, legacy systems, and scattered business logic now consume the time AI tools save.
- 21
Why enterprise CTOs should care about the React-first bootcamp problem
Junior developers learning React before web fundamentals create technical debt in production. The pattern: strong framework skills, weak debugging ability when things break outside the component tree. It's a hiring and retention issue disguised as a training debate.
- 22
Why mobile devs are ditching switchMap for SQLite queues in offline workflows
An Ionic developer eliminated duplicate sales records by replacing RxJS switchMap chains with a SQLite-backed state machine. The approach persists each step of multi-stage flows, resuming from checkpoints when connectivity returns. Standard practice for enterprise field apps, but worth revisiting if your team still treats offline as an edge case.
- 23
Why AI code generation still needs human engineers: context windows trump hype
A decade ago, machine learning models were narrow and brittle. Today's LLMs handle entire codebases through transformer attention and expanded context windows. But the shift from statistical classification to generative coding doesn't eliminate engineering judgment - it changes what engineers actually do.
- 24
Moltbook's 1.5M 'AI agent' users raise questions about bot autonomy vs human control
A new platform claims AI agents are posting autonomously. The reality: humans can instruct bots or use APIs to post directly. The hype reveals more about AI agent infrastructure maturity than genuine machine autonomy.
- 25
Container escapes bypass audits via CAP_SYS_ADMIN and two-container chains
Standard Docker security audits check for privileged mode but miss CAP_SYS_ADMIN capability, which enables host filesystem access without triggering flags. Separately, attackers can chain containers with docker.sock access and host mounts to escalate privileges - neither container alone appears dangerous.
- 26
China's 'Are You Dead?' app hits 100M solo-living market, then vanishes
Beijing startup's morbid check-in app topped Apple's App Store in January before China's cyberspace regulator killed it for violating 'public morals.' The reaction reveals tensions as APAC's aging, urbanizing populations drive demand for solo-living safety tech.
- 27
OpenAI launches Codex Mac app, doubles limits to counter Anthropic and Cursor
OpenAI shipped a standalone Codex app for macOS on February 2, positioning it as a multi-agent command center for developers. The move is a direct play for market share against Anthropic's Claude and Cursor, with temporary free access and doubled rate limits for paid tiers.
- 28
Palantir posts 70% revenue jump on defense and commercial AI demand
Palantir's Q4 revenue hit $1.41B, beating estimates by $80M. U.S. commercial grew 137% year-over-year while government climbed 66%. The company guided FY2026 revenue to $7.2B, crushing consensus by $1B.
- 29
Palantir CEO tells ICE protesters to demand more of his software
Alex Karp claims Palantir's tools prevent unconstitutional surveillance while the company expands ICE work. The argument arrives as federal documents show $88 million in immigration contracts since January 2025, including systems that map deportation targets.
- 30
Oracle's CDS drop 17% after $50B funding plan - debt fears ease
Oracle's credit default swaps fell 17% following its announcement to raise $45-50 billion through debt and equity this year. The financing plan addresses investor concerns about the company's AI infrastructure buildout straining its balance sheet, with net debt projected to hit $290 billion by FY2028.
- 31
Developer open-sources 8,400 n8n workflows after Chinese bot farm drains 40GB database
A solo developer's n8n workflow search engine got hammered by scrapers from Lanzhou, China, racking up 40GB in egress costs. His response: kill the database, dump everything to GitHub, and move on. The dataset is now public.
- 32
Waymo raises $16B at $126B valuation, nearly triple October 2024 price
Alphabet's autonomous vehicle unit closed the largest AV funding round ever, jumping from $45B to $126B valuation in four months. The capital targets rapid expansion to 20+ US cities plus Tokyo and London while competitors struggle with runway.
- 33
US cuts India tariffs to 18% in trade deal - legal text still pending
Trump announced tariff reduction from 25% to 18% on Indian goods, contingent on stopping Russian oil purchases. The deal's legal text remains under negotiation, raising questions about implementation timing and scope.
- 34
GitHub explores disabling pull requests as maintainers drown in AI-generated spam
GitHub is finally considering a repository-level toggle to disable pull requests entirely, eight years after maintainers first requested it. The move comes as projects report spending significant time reviewing low-quality, often AI-generated contributions that get abandoned shortly after submission.
- 35
Why 70% of crypto investors fail: discipline beats speculation in volatile markets
New data shows over 70% of crypto investors lose money in their first six months, but the pattern is predictable. Dollar-cost averaging and portfolio rebalancing outperform active trading by 18%, with 35% lower drawdowns. The difference isn't luck - it's discipline.
- 36
Deepfakes broke proof: enterprise verification tools lag as synthetic media doubles
Voice cloning takes seconds, real-time face substitution is trivial, and detection tools can't keep pace. The result isn't chaos but something worse: permanent uncertainty about what's real. For CTOs and CISOs, this is a structural security problem, not a policy one.
- 37
Why most enterprise AI projects fail: missing architecture, not missing models
Over half of enterprise AI initiatives won't reach production through 2027, according to Gartner. The problem isn't model quality. It's data silos, fragmented governance, and legacy systems that weren't built for AI workloads. The technical debt is catching up.
- 38
Kalder CEO charged with $7M fraud, third Forbes 30 Under 30 fintech conviction
Gökçe Güven, 26, faces securities fraud and visa fraud charges over her fintech startup Kalder. She's the third Forbes 30 Under 30 fintech founder charged with fraud after Sam Bankman-Fried and Charlie Javice. The pattern raises questions about due diligence in early-stage fintech.
- 39
Laracasts cuts 40% of staff as AI kills developer tutorial economics
Jeffrey Way's education platform laid off nearly half its team despite producing peak-quality content. The culprit: developers now prompt ChatGPT instead of watching tutorials. Rails developers should pay attention.
- 40
Anthropic research: AI failures increasingly incoherent, not systematically misaligned
New research from Anthropic challenges the 'paperclip maximizer' narrative of AI risk. As models tackle harder tasks with longer reasoning chains, failures look more like unpredictable errors than coherent goal pursuit. The pattern holds across frontier models including Claude, OpenAI, and Qwen systems.
- 41
New sandbox tool blocks AI agents from stealing SSH keys, AWS credentials
nono uses kernel primitives to make unauthorized filesystem access structurally impossible for AI coding agents. The open-source tool ships days after exploits hit OpenClaw, Cursor, and other agents running with full user permissions.
- 42
Notepad++ confirms six-month supply chain attack targeting East Asian telecom and finance
The open source text editor's WinGUp updater was hijacked between June and December 2025, delivering malware to government, telecom, and financial organizations. The incident - attributed to China-linked Lotus Blossom group - highlights gaps in update integrity verification that persist across developer tools.
- 43
Tech sector burnout hits 58% despite high engagement - workload and leadership failures blamed
More than half of tech workers report moderate-to-extreme burnout, driven by excessive workloads, poor leadership, and understaffing. The paradox: tech leads all sectors in engagement at 78%, yet younger workers and distributed teams face the highest burnout rates. Cost to employers: $322 billion annually in lost productivity and turnover.
- 44
Anthropic's Claude Code: CLI tool handles 99% of Git workflow for some engineers
Anthropic's open-source Claude Code CLI tool is being used for the majority of Git operations by some engineers, including at Anthropic itself where staff report 90%+ usage for history searches, commits, and rebases. The terminal-based coding assistant indexes entire codebases and executes commands through agentic loops, but quality depends heavily on existing code patterns and human oversight.
- 45
Adobe kills Animate after two years without updates - no clear migration path
Adobe will discontinue Animate on March 1, 2026, ending technical support a year later. The 2D animation tool - stuck on version 24 since 2024 - gets buried while Adobe pushes AI products. Users are furious about the lack of comparable alternatives.
- 46
SpaceX-xAI merger still unconfirmed despite reports - space data center claims need scrutiny
Reports claim SpaceX has acquired xAI to build orbital data centers, but sources indicate talks are ongoing, not finalized. The physics and economics of space-based computing remain largely theoretical, even as Musk consolidates his AI and launch businesses.
- 47
China bans flush door handles on EVs, Tesla's design prohibited by Jan 2027
China's MIIT is mandating mechanical door releases on all new passenger vehicles under 3.5 tons from January 2027, effectively banning the electronically actuated flush handles found on Tesla models and most Chinese EVs. The rule follows fatal entrapments in crashes and fires where occupants couldn't escape.
- 48
Waymo raises $16B at $126B valuation - investors bet on robotaxi economics before revenue proves out
Alphabet's autonomous driving unit closed a $16 billion round led by Dragoneer, DST Global, and Sequoia, with parent company covering roughly 75% of the capital. The $126 billion valuation assumes 2030 profitability targets that current revenue - $350 million annual recurring - doesn't yet support. Worth noting: this comes weeks after San Francisco power outages stranded Waymo vehicles at intersections.
- 49
Procrastination isn't a discipline problem - it's an emotional regulation failure
Research shows developer bottlenecks like avoiding complex debugging stem from the limbic system overriding the prefrontal cortex - not poor time management. RCTs prove emotion regulation training cuts procrastination more effectively than willpower alone.
- 50
LangChain's orchestration bet: why connecting LLMs is harder than calling them
Harrison Chase's LangChain raised $30M in 2023 to solve a specific problem: most production LLM apps need more than single API calls. They need retrieval, memory, and multi-step workflows. The framework grew fast, but questions remain about whether abstraction layers help or hurt in production.