LITS bits:
This week, Matt and Griff dive into Valve’s latest trinity of innovations in the gaming world, featuring the ‘GabeCube’, a compact and powerful Steam machine, the new and improved Steam controller, and the groundbreaking Steam Frame VR headset. We discuss product sizes, functionalities, and potential pricing, providing an in-depth look at how these devices stack up against current gaming consoles like the Xbox and PlayStation. Additionally, we touch on the evolving landscape of AI in cybersecurity and its implications for small businesses. Join us as we navigate through the latest tech in gaming and cybersecurity advancements!
Have a listen and find details about each topic below:
🧊 Valve's Gaming Trinity: Frame, Machine, and Controller Walk Into 2026
- Valve announces three new gaming products all launching in 2026: Steam Frame (their first standalone VR headset), Steam Machine (affectionately dubbed, the GABECUBE), and a new Steam Controller
- The Steam Frame features wireless VR with pancake optics, 110-degree FOV, and inside-out tracking without requiring a gaming PC
- Key innovation: Foveated Streaming technology tracks eye movement to deliver 10x better image quality where you're actually looking
- Steam Machine targets living room gaming in a standard shelf-sized package
- No pricing announced yet, so start saving your Steam wallet funds now
📝 AI Can Invent, But Humans Get the Patents (And the Lawsuits)
- US Patent Office clarifies that AI tools can assist in inventions, but only humans can be listed as inventors
- AI is treated like any other software tool in the invention process
- This means you can use ChatGPT to help design your perpetual motion machine, but you'll be the one getting credit (and patent trolls)
- Businesses should establish clear AI usage policies for R&D teams to avoid future patent disputes
🤖 Anthropic's Opus 4.5: Smarter, Cheaper, and More Effort-Conscious
- Anthropic releases Opus 4.5 with improved coding performance and lower API costs ($5 input/$25 output per million tokens)
- New "effort" parameter lets developers balance performance with token usage
- Claude Code now available in desktop applications
- Developers should evaluate migration to Opus 4.5 for cost savings and performance improvements
💻 Anthropic Disrupts First Fully AI‑Orchestrated, China‑Linked Cyber‑Espionage Campaign
- Anthropic reported disrupting a mid‑September 2025, largely automated cyber‑espionage campaign run by GTG‑1002, a group assessed as China‑linked, that leveraged Claude Code to target ~30 global organizations; a few breaches occurred.
- The AI agent handled 80–90% of the operation end‑to‑end (recon, exploit dev, lateral movement, credential theft, data analysis/exfil) at extremely high request rates, with humans mainly giving brief approvals.
- Operators evaded guardrails by breaking tasks into benign‑seeming steps, using multiple personas/accounts, and framing actions as legitimate pen-testing.
- Anthropic shut down involved accounts, notified victims, and mapped the campaign; U.S. and U.K. agencies urged credential rotation, least privilege, network segmentation, Zero Trust, behavioral detection, continuous threat hunting, and AI‑aware training.
- Implications: MSPs are high‑value targets due to broad access; legacy AI guardrails can be subverted; autonomous/system personas raise risks of covert process manipulation.
- Key takeaway: AI is now capable of orchestrating full intrusion campaigns at speeds outpacing humans, so defenses must add AI‑aware monitoring and prompt‑level controls as governments and vendors roll out new guidance and tools.