@miranetwork profile picture
Mira @miranetwork

Building the Trust Layer for AI.

Joined February 2024
  • Tweets 5,557
  • Following 3
  • Followers 250,963
1,321 Photos and videos
Pinned Tweet
Mira
@miranetwork
26 Sep 2025
Mira Mainnet is Live. The trust layer for AI has arrived.
1:35
817
578
2,686
1,101,714
What dodging AI hallucinations and its unverified outputs does to a man.
7
14
1,372
Humans have biases. Data has biases. Models learn from data. The math was never the hard part when figuring out AI.
0:05
19
2
22
1,830
Asked an AI model to identify a song from a reel the other day. It suggested me to use Shazam instead. Wdym, you'll give me unverified responses only on important and intellectual stuff!
0:06
12
19
1,659
Verify
28
1
50
3,092
Trust me, bro, I'm not dependent on AI! Me when the subscription ends :
0:12
14
28
3,101
Watching the AI give me the same response for the 11th time after religiously ignoring my instructions
0:05
18
25
2,201
Ozempic to get rid of fat? How about you try to get rid of AI hallucinations?
12
19
3,508
What would you do if AI stopped hallucinating for a Day?
16
18
2,705
4 yr Olds are smarter than AI sometimes
16
2
28
3,136
We got God on Claude before GTA 6.
Kristof
@CoastalFuturist
May 25
BREAKING NEWS: God joins Anthropic as member of technical staff
10
23
3,501
neat af
Solana
@solana
May 12
22
5
44
5,008
QT with the darkest AI conspiracy theory you know !!
26
2
29
6,767
we didn't stop driving cars when people started crashing we added seatbelts verification is the seatbelt for AI agents
31
50
8,608
stages of agent deployment: 1. "it works perfectly in testing" 2. "why did it do that" 3. "how much did we lose" 4. "maybe we should've verified first" 5. acceptance
21
2
41
5,757
current state of agent security: "we tested it a few times and it seemed fine" that's not a security model. that's vibes.
19
3
25
5,819
2024: "AI can write essays" 2025: "AI can write code" 2026: "AI can manage your money" 2027: "AI managed your money"
23
5
44
4,474
agents making payments: solved agents holding wallets: solved agents getting identity: solved agents proving their outputs are correct: ← hi
35
3
53
5,910
an AI agent making a wrong API call = annoying an AI agent sending a wrong email = embarrassing an AI agent executing a wrong transaction = expensive
27
3
57
6,684
what's the max amount you'd let an AI agent control? $10? $1,000? $100,000?
43% $10
32% $1000
25% $10000
150 votes • Final results
28
2
51
7,697
"move fast and break things" was fine when you were breaking UI layouts it's less fine when you're breaking someone's treasury
22
2
65
5,661
Load more