Anduril has developed the X-47 collaborative combat aircraft (called Fury or FQ-44), an autonomous fighter jet that flies alongside manned fighters, extending their weapons and sensor reach while engaging in high-risk missions, and achieved first flight in 556 days from contract signature—the fastest new fighter development since the Korean War.

Anduril operates on a product-company model rather than a defense-contractor model, investing its own capital to build products it believes are right and then selling them to the government, rather than accepting government contracts to perform work.

The product-company model aligns incentives toward speed and quality: the company makes more money when it moves faster rather than slower, and makes more money when things work the first time rather than taking multiple attempts, which is the opposite incentive structure of cost-plus contracting.

Optical camouflage in the visible spectrum is technically achievable through projection systems that calculate the luminance and color values needed to match the background, and Anduril has demonstrated this on its Ghost drone.

UAPs are more likely to come from Earth's past than the future, because there has been far more time in Earth's past (hundreds of millions of billions of years) than present, making it statistically more probable that advanced entities come from the past, and forward time travel through gravity distortion is more physically plausible than backward time travel.

Unidentified aerial phenomena (UAPs/UFOs) may be difficult to detect on radar because they might not interact with physical reality in the way we expect, and some appear to have no radar signature despite being visible, suggesting either active cloaking or a different mode of physical existence.

Optical camouflage is relatively easy and not that militarily useful because modern adversaries (Russia, China, Iran) use thermal, radar, lidar and other sensing technologies that easily defeat visible-spectrum cloaking, making optical camouflage mainly useful only against adversaries with naked-eye observation only.

Ukraine's Decentralized Drone Production Is Double-Edged

Ukraine's enormous, decentralized drone production taps a whole-of-society ecosystem of engineers and entrepreneurs building drones in garages with Chinese components and 3D printers, which enables scale and speed but also produces chaotic, non-standardized solutions because it isn't a top-down organized effort.

Anti-Personnel Autonomy Is The Most Bannable Case

International bans against autonomous weapons that target people may be more achievable than broad bans because their military utility is lower (humans don't move fast, so keeping a human in the loop is feasible), the moral 'ick factor' is higher, and the moral arguments are stronger—since a soldier can flee a tank or surrender to avoid being targeted but cannot stop being a human.

Humans Stay Near Front To Quarterback Drones

Humans will likely remain physically at risk near (not on) the front lines because jam-resistant long-range communications are extremely hard while short-range protected links are achievable, so operators must stay close to 'quarterback' their drone swarms—becoming high-value targets themselves, since neutralizing the human neutralizes the swarm.

If Soldiers Are Unreachable Civilians Become Targets

As soldiers leave the physical battlefield, civilians in the rear may increasingly become targets, continuing a historical pattern in which militaries deliberately target civilian populations to cause suffering and break societies—a tactic that is illegal under the law of war but recurrently used.

Emitting Targets Are Easiest For Autonomous Weapons

Targets that emit in the electromagnetic spectrum (radars on air-defense systems or ships) are the easiest for autonomous weapons because their signals can be characterized in peacetime and programmed in; this technology dates to the 1980s, as with the US Tomahawk anti-ship missile and the Israeli Harpy drone.

Drone Swarming Means Autonomous Cooperative Behavior

True swarming is not merely many drones attacking at once but autonomous cooperative behavior—drones communicating and coordinating to mount adaptive multi-axis attacks, disperse and counterattack against enemy movement, and form self-healing communication networks or minefields—which could unlock transformative effectiveness unlike human coordination, but this is a longer-term proposition not yet realized.

Military AI Inevitable But Foolish Use Is Not

Militaries will certainly use AI for command and control, information processing, and better decisions—that much is inevitable—but doing so in a bad, foolish, or dangerous way is not inevitable, leaving room to craft rules that bound how nations employ AI and autonomy.

Terrorists Prefer Guns And Vehicles Over Drones

We haven't seen widespread terrorist use of attack drones largely because someone wanting to cause mass harm can kill far more people more easily with an AR-15, a handgun, or even a vehicle ramming attack, all of which are already widely accessible—making engineered drone attacks comparatively painstaking and less lethal.

Nominal Human In The Loop As Scapegoat

Even with a human nominally in the loop—a person pushing a button to approve attacks—they may have effectively ceded all judgment to the machine, serving merely as a scapegoat for accountability rather than exercising meaningful control, which is both morally and potentially legally unacceptable.

War Is Terrible And Should Not Be Sugarcoated

War is terrible and we should not sugarcoat it; we should focus on weapons that are effective even though they are horrible, because that is the reality of war.

Autonomous Weapon Defined As Machine Choosing Targets

An autonomous weapon is conceptually simple: it is a machine making the decision on the battlefield about who to kill; the complexity only arises when examining how automation is incrementally applied across different functions of warfare.

Air And Missile Defense Already Fully Automatic

At least 30 countries already operate air and missile defense systems with fully automatic modes that, once activated by a human, will shoot down incoming missiles, rockets, or artillery shells on their own if the threat meets pre-programmed characteristics—but this is a narrowly bounded domain.

Self-Driving Car Analogy For Incremental Weapon Autonomy

Just as cars added automation incrementally (cruise control, automatic braking, self-parking, full self-driving), militaries are incrementally adding automation to distinct engagement functions—target identification, target validation, weapon selection, attack decision—rather than fielding a single all-in-one autonomous weapon.