Nội dung
- Zeta partners with Palantir to build AI marketing infrastructure for enterprises
- An old video in Iran continues to highlight the pressure to verify content in the AI era
- ByteDance unveils Seedance 2.5, raising the bar for AI video with longer and sharper output
- Visa card, self-custody, and Tria’s cross-border payment experience
- Norway wants to nearly completely restrict AI for younger students
- Canopy pushes an AI-native appchain model with an “Own The Controls” approach
- Picsart and Seedance 2.0 continue to prove the appeal of prompt-based video creation tools
- Market volatility and the AI infrastructure race continue to heat up
- Next-generation AI models continue to pressure products relying on “thinking tokens”
- What this means for the Vietnamese market
- References
This latest batch of AI news shows how quickly technology is shifting across three layers: enterprise infrastructure, creative tools, and regulatory policy. For Vietnamese marketers, it is not just about what AI can do, but also about how platforms, data, copyright, and content distribution will change over the next year.
Zeta partners with Palantir to build AI marketing infrastructure for enterprises
Zeta and Palantir have just announced a strategic partnership to build AI marketing infrastructure on the Palantir Foundry platform. According to the information shared, Palantir will also bring Zeta’s AI marketing tools to eligible Foundry customers, opening up a revenue opportunity that could exceed $100 million annually over time.
The notable point in this deal is not a single feature, but the fact that AI marketing is moving closer to an enterprise infrastructure model. For large brands, the advantage will go to systems that can connect data, operations, and campaign activation on the same technology layer.
Source: Shay Boloor (@StockSavvyShay)
An old video in Iran continues to highlight the pressure to verify content in the AI era
An Iranian math teacher posted a new video to prove that a previous clip was not fake or AI-generated. Although the source information is brief, the situation reflects a growing reality: as generative tools become more powerful, viewers tend to question every image and video online.
For communications professionals, this is a reminder that content trust is becoming more valuable than ever. It is no longer enough to create quickly; brands also need ways to prove the authenticity, source, and context of the content they publish.
Source: Iran in English (@Iran_in_Eng)
ByteDance unveils Seedance 2.5, raising the bar for AI video with longer and sharper output
ByteDance has officially introduced Seedance 2.5 along with the Seedance 2.0 update. According to the source, Seedance 2.0 already supports 4K video output, while Seedance 2.5 can generate 30-second videos in a single generation. The company also announced a new AI copyright commercialization platform.
From a content-market perspective, this is an important signal because technical barriers are falling quickly: AI video is no longer just for testing short ideas, but is moving toward being long enough, sharp enough, and stable enough for commercial campaigns. The addition of a copyright platform also shows that the AI ecosystem will increasingly be tied to lawful IP usage.
Source: TestingCatalog / quoted from @xiaohu
Visa card, self-custody, and Tria’s cross-border payment experience
Tria is being cited as an example of the trend toward simplifying the crypto experience: users still keep their private keys in their wallet, but can spend in stores, pay for services, send international transfers, and use multiple tokens across multiple chains. According to the source description, the BestPath system handles routing behind the scenes, with no bridge, no gas, and no wrapped token; the Visa card works in more than 150 countries and offers up to 6% cashback.
The key point for product builders is that the challenge in AI and Web3 is not only about “powerful technology,” but also about “is it usable?” If automation infrastructure removes the need for users to understand the technical side, adoption will rise significantly.
Source: OG FIGO (@OGFIGO)
Norway wants to nearly completely restrict AI for younger students
According to the information cited, Norway has announced plans to almost completely ban AI for younger students. Prime Minister Jonas Gahr Stoere was quoted as saying that the most important thing in school is for children to learn reading, writing, and arithmetic.
This move creates an interesting counterpoint to the wave of early AI adoption in education in many other countries. For regulators and edtech companies, the question is no longer whether to use AI, but where, at what age, and for what pedagogical purpose.
Source: Dexerto
Canopy pushes an AI-native appchain model with an “Own The Controls” approach
Canopy is drawing attention with its “Own The Controls” philosophy and an AI-native framework for deploying sovereign appchains. According to the source, the project has opened a creator leaderboard on Nucleus Codes with $100,000 in $CNPY for the top 300 contributors, while also reporting that its testnet has seen thousands of chains created and millions of transactions.
The project also says it has added templates supporting Python, TypeScript, and other languages familiar to developers, along with a community ownership mechanism for supply, validator locks, and network security. This is a sign that AI-Web3 infrastructure is competing by attracting builders into the ecosystem, rather than relying only on technology messaging.
Source: H B O (@HBO_data)
Picsart and Seedance 2.0 continue to prove the appeal of prompt-based video creation tools
A creative example shared on Picsart shows users leveraging GPT Image 2 and Seedance 2.0 to create dramatic scenes, from player imagery to a virtual tiger effect in the background. Although this is demo content, it clearly reflects market demand: turning ideas into moving visuals using only text descriptions.
For marketers, the lesson is that tools like these are opening up new ways to produce short ads, product teasers, and social content. However, to use them effectively, brands still need tight control over character consistency, setting, and messaging.
Source: Sharon Riley (@Just_sharon7)
Market volatility and the AI infrastructure race continue to heat up
Today’s market roundup shows that many capital and AI infrastructure flows are still accelerating. The source mentions that a potential end-of-quarter rebalancing could lead institutional investors to sell up to $165 billion in stocks and move a corresponding amount into bonds; meanwhile, SpaceX is said to have signed a major compute deal with Reflection AI, gaining access to Nvidia GB300 chips at Colossus 2, while large-cap ETF groups also recorded about $150 billion in retail inflows over the past month.
For the AI industry, this shows that compute capacity remains a strategic asset. Models, startups, and creative platforms are all competing not only on algorithms, but on access to large-scale, stable compute.
Source: amit (@amitisinvesting)
Next-generation AI models continue to pressure products relying on “thinking tokens”
According to the shared view, GPT 5.6 is expected to be cheaper, faster, and more practical, which could erode the advantage of products like Fable. The comment also points out that Anthropic models consume many tokens for “thinking,” increasing costs and reducing competitiveness in certain real-world use cases.
Although this is an opinionated take, it reflects a very clear trend in the AI market: customers are increasingly focused on speed, cost, and output efficiency rather than just model capability specs. For Vietnamese businesses, this is especially important when evaluating AI tools for content, customer support, or operational automation.
Source: Bindu Reddy (@bindureddy)
What this means for the Vietnamese market
Overall, these 9 stories show that AI is moving in three main directions that Vietnamese businesses should closely follow. First, AI is becoming tied to data infrastructure and real operations, as seen in how Zeta and Palantir are approaching enterprise marketing. Second, generative AI is entering a deeper commercialization phase, where video, copyright, and IP are becoming core issues rather than just image quality.
Third, competition in AI will increasingly revolve around cost, speed, and the ability to integrate into existing workflows. For Vietnamese marketers, this means AI should not be tested only as a “fast content creation” tool, but evaluated as part of a system: input data, approval workflows, content usage rights, and the ability to measure effectiveness.
From a governance perspective, Norway’s move is a reminder that not every context should open AI up to the maximum. In education, finance, healthcare, or other high-risk sectors, Vietnamese businesses will need clearer principles for control, verification, and access permissions when using AI.
Ultimately, today’s AI wave is shifting from “trying something new” to “choosing what fits.” Businesses that define their goals, data, and usage boundaries early will get more value from AI than those simply chasing the trend.
Compiled from the international sources listed below.
See more marketing news and guides at https://marketing365.vn.
Follow more updates from Marketing365 to stay up to date with the latest marketing trends.
Read more articles in the same category AI Developments.
This article focuses on new AI news with a perspective for the Vietnamese market.
References
- Shay Boloor (@StockSavvyShay)
- Iran in English (@Iran_in_Eng)
- TestingCatalog (@testingcatalog)
- OG FIGO (@OGFIGO)
- Dexerto
- H B O (@HBO_data)
- Sharon Riley (@Just_sharon7)
- amit (@amitisinvesting)
- Bindu Reddy (@bindureddy)
