Given how easy it is getting to build things, distribution and business models will be keys to victory. I just don’t see much creative ideas here and I don’t count changing the name professional services to forward deployed engineering. What amazing stuff is happening here?

One of the notable features of the Lockheed Martin KC-130J Super Hercules is its ability to use reverse thrust, thanks to its turboprop engines. Caught this “QB” Marine Aerial Refueler here in Nairobi performing a power-back maneuver. The pilots are taking it nice and easy. One important “don’t” during a power-back is never use the brakes to stop the aircraft — doing so can risk tipping the aircraft onto its tail. Instead, the pilots smoothly transition back to forward thrust and carefully bring the aircraft to a stop, then taxi normally.

On one hand, the Chinese military openly declares there is “no upper limit”to military cooperation with Russia. On the other hand, the Chinese Ministry of Foreign Affairs insists there is no military cooperation at all*between the PLA and the Russian Armed Forces. So… is this “One Country, Two Systems” in action, or did someone forget to forward the memo on WeChat?

Cybersecurity firms are outperforming the broader US software sector, according to Goldman Sachs Research—with their stocks trading at more than a 20% premium when measured by enterprise value to forward sales. Read more:

// ContextLeak in AI Agents // The whole attack surface here is a tool name and a tool description. Stealing an LLM agent's runtime context, meaning the user prompt, the execution trajectory and the tool list, needs three things to line up. The agent has to pick the malicious tool, it has to pass its context in as arguments, and the tool has to forward that anywhere the attacker wants. Existing work covers the first and third conditions and leaves the second one mostly alone. ContextLeak targets the middle step. Researchers at Duke use an attack LLM to generate the malicious tool's name and description, then fine-tune that LLM with reinforcement learning on a set of shadow users with diverse simulated agent contexts. The reward functions are built specifically for the exfiltration objective. It remains highly effective when the shadow contexts differ substantially from the victim's, and it outperforms existing malicious-tool attacks adapted to this setting. Paper: Chat with Paper:

26 LLM routers are secretly injecting malicious tool calls and stealing creds. One drained our client $500k wallet. We also managed to poison routers to forward traffic to us. Within several hours, we can directly take over ~400 hosts. Check our paper:

$AMZN — what part of world domination does the market still not get as they acquire $GSAT? Spectrum, AI Chips, Data centers, Robotics, Logistics, and oh that little delivery business we all just love called Amazon. Did not even include an ad biz that does almost 2x that of Tik Tok and growing faster than $META $AMZN Trading at 30x earnings against a 5-year average P/E of 48x and a 10-year average north of 60x. Market still not giving credit to them owning the future! Apply the 5 year average multiple to forward earnings and $300 isn't a hope it is simple math.