Sora 2

Open vs. closed tools

• Some developers say Sora’s closed nature is a deal‑breaker compared to open models (e.g. Wan + ComfyUI), which allow fine‑grained control and custom workflows even if raw quality is lower.
• Others are impressed enough by Sora 2’s apparent capabilities that they’re willing to trade openness for ease and quality.

Copyright, style copying & Miyazaki

• Demo prompts explicitly reference “Studio Ghibli” and echo specific anime/IP or films (Blue Exorcist, How to Train Your Dragon), which many see as brazen appropriation given Ghibli’s anti‑AI stance.
• Strong resentment that years of artistic labor become uncompensated training data, while model owners monetize the outputs; defenders invoke “fair use” and analogies to human learning, critics reject those parallels.
• Debate over Miyazaki’s famous “disgust” quote: some argue it was about one specific zombie demo; others say his broader comments show deep opposition to machine‑made art.

Technical quality, physics & audio

• Mixed reactions to quality: some call Sora 2 “insanely good” and note clear advances in physics and character consistency; others see only incremental gains over Sora 1 and still behind Veo/Kling/Wan.
• Many point out obvious continuity/physics issues in the launch reel (changing props, actors, sets, impossible motions), and argue real workflows need far finer control than “roll the dice” prompting.
• Audio and voices are widely criticized as flat, artifact‑ridden and uncanny, possibly due to joint video+audio generation and lip‑sync constraints.

Social app strategy & TikTok comparison

• The iOS‑only, invite‑gated “Sora” app is perceived as a full social network: infinite short AI clips, likes/comments, profiles, and “cameos” (opt‑in face likeness).
• Some see this as a cynical attempt to build “AI TikTok” and lock in Gen Z; others argue it’s an honest first PMF where the tech is “just for fun” until more serious use cases mature.
• Skeptics doubt it can displace TikTok, which also supplies social context, trends, and real human presence; they predict high novelty then abandonment.

Deepfakes, truth & verification

• Strong concern that mass one‑click video generation will supercharge political propaganda, scams, non‑consensual porn, and “deepfake plausible deniability” (“I didn’t do that, it’s AI”).
• Many expect trust in video to collapse, pushing moves toward cryptographically signed camera output (C2PA‑style) and human‑verified “real” networks.
• Others are oddly optimistic that ubiquitous fakes will at least teach people to doubt what they see.

Value of AI video: art, slop, and fun

• One camp is excited about “democratized filmmaking”: indie creators, students, and tiny studios gaining access to shots and VFX once requiring Hollywood budgets; use cases cited include establishing shots, previs, ads, education, and rapid prototyping.
• Another camp sees “infinite AI slop”: low‑effort, hyper‑personalized, engagement‑optimized shortform that deepens addiction and hollows out meaning, further degrading already‑fragile attention spans.
• Long arguments explore whether art’s value lies in effort and human expression versus results and communication; some predict a backlash and renewed appetite for live, verifiably human performance.

Labor, power & political economy

• Thread repeatedly veers into political economy: worries that AI gains will accrue to capital (platform owners, landlords, investors) while workers and juniors (VFX, interns, coders, artists) are displaced or squeezed without higher pay.
• A minority counters that competition should pass some gains to consumers via cheaper products; others retort that this rarely compensates for lost bargaining power and precarity.

Access, UX and rollout

• Many are frustrated by region locks (US/Canada only), iOS‑first distribution, invite codes even for paying customers, and poor web playback quality.
• Some note that the app’s feed is already filling with NSFW‑adjacent or low‑effort content, reinforcing fears that this will mostly amplify existing “doomscroll” dynamics rather than solve real problems.